diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/record.json new file mode 100644 index 0000000000000000000000000000000000000000..d01cebebeaad1e1a1e5a4ccec38894bb55308c5c --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d3_17f4b0cf2f62", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic bulletin board with a green marker on top of similar markers and colored pencils, high contrast and clear separation", + "target_evidence": "green marker", + "expected_visual_elements": [ + "bulletin board", + "markers", + "colored pencils" + ], + "negative_prompt": "low resolution, overlapping elements, watermarks", + "diversity_tags": [ + "color similarity", + "object placement" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "17f4b0cf2f62", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic bulletin board with", + "max_similarity_to_other_prompt": 0.48, + "diversity_score": 0.32, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_17f4b0cf2f62", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24786734, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 565084, + "qwen_elapsed_seconds": 5.099, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..d577b9610c9f53fae20ea403c92554b7fade4b73 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a1a93d0192ed79c42321ead54248679004e7d9047764c18a03e7083778267dd +size 551822 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..70280f050e89615885148fe89de7aa9b78a8f319 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4899cde139d67894179c203f7099a30cab7466ec3dc3ca34e547d9afe36144ff +size 581955 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..622538435f8ce68e8a7ebef7ea0b92fa67f08fae --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7c740bfd03fdb77f856788c32db7d98778384e873fd4d4aa7070446096d0a7f5 +size 568370 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/record.json new file mode 100644 index 0000000000000000000000000000000000000000..ab747a1ba5841b1ec3e81406b0eed0cc38f2bafe --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "clutter_distractor_d3_1fb8fe8d0e70", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic shelf with a blue bottle standing out among glasses, cups, and jars, clear boundaries for easy detection", + "target_evidence": "blue bottle", + "expected_visual_elements": [ + "bottle", + "glasses", + "cups", + "jars" + ], + "negative_prompt": "overlapping items, low contrast, out-of-focus", + "diversity_tags": [ + "color similarity", + "object scale" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "1fb8fe8d0e70", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic shelf with a", + "max_similarity_to_other_prompt": 0.3214, + "diversity_score": 0.4786, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_1fb8fe8d0e70", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25913680, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 3514110, + "qwen_elapsed_seconds": 5.164, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..67e993722a847d8aff46587b14f7373c05c430b6 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0359ecd142afd68701d059ec3b82fa08728b9c4b7d6a3a5c2a10b0ac06634b45 +size 337894 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..de9543c2a279bdad880a2bfb3b9511cbf88f1efd --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c41a1833d13e59a3e4ef905e4171fe5521f975f006003d154874c83a1cbb8b05 +size 662462 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..735b26a3a547de8632f00914beaa15275ade3576 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:46aeb9745db5f9aa86f71f19059f439aebbf556732c103ff7cfc924cbbf289f3 +size 578487 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/record.json new file mode 100644 index 0000000000000000000000000000000000000000..20d2886dbb5319a54a9a4746e14904909c13e8a8 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "clutter_distractor_d3_24fe198c21d6", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic workbench with a blue screwdriver surrounded by nuts, bolts, and wires, sharp focus and clear foreground separation", + "target_evidence": "blue screwdriver", + "expected_visual_elements": [ + "workbench", + "nuts", + "bolts", + "wires" + ], + "negative_prompt": "low resolution, cluttered background, watermarks", + "diversity_tags": [ + "shape similarity", + "layout variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "24fe198c21d6", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic workbench with a", + "max_similarity_to_other_prompt": 0.52, + "diversity_score": 0.28, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_24fe198c21d6", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_24fe198c21d6/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25933561, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2257078, + "qwen_elapsed_seconds": 5.197, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..fbfd2c2ae37a613d8d6166f5a137738bd81d874e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:12396d5366a977d76d9d2245bdb51cc6b9ec2e9a990c514c6ea8ebc3f7bed2bb +size 340885 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..5e7130f3a20a9c2395ce85697d489a58b6291673 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:403223d089315ae36964a588c054088d4bec9b2b31aa5793d8e0b83fc7552e7a +size 817223 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..45cd67e5d420cd1d91f62a5c7e19625cc06864a9 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:564688bd399ba22e1074de25b65ebdeccc180ffe6e17ba119ca4b4497fc6dbdf +size 548249 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/record.json new file mode 100644 index 0000000000000000000000000000000000000000..0356a796645a75b185445f1fdeabe20c1da5cced --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "clutter_distractor_d3_2a92bf5e61f6", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic bulletin board with a yellow sticky note among other notes, paper scraps, and markers, clean foreground separation", + "target_evidence": "yellow sticky note", + "expected_visual_elements": [ + "sticky note", + "paper scraps", + "markers", + "post-it notes" + ], + "negative_prompt": "blurred edges, mixed colors, overlapping notes", + "diversity_tags": [ + "shape similarity", + "text region" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "2a92bf5e61f6", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic bulletin board with", + "max_similarity_to_other_prompt": 0.64, + "diversity_score": 0.16, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_2a92bf5e61f6", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20905239, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 8752243, + "qwen_elapsed_seconds": 5.124, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..eec4880c054afec64f19b7dae4638c03781e5c7c --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9413a55296d520cadf8a3a389af21263091233db5654a4f51a8426d1058e40e5 +size 482459 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..5beb36a635f8d946ee23c9c76dc421136f63fa7b --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e8a4bd39a031f22bbf2a4a59e9cfaf69bbd46d93bade2e23b5ac291266acb9f +size 483432 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..89d4064580c45be476efed838a7536b610398395 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8cc49fa5ee154baa9f92d76ebc8b64a50602b5a72e4d1fc42cc29e32edc90ffb +size 504764 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/record.json new file mode 100644 index 0000000000000000000000000000000000000000..11e1f33700b944047ada49e262a83675ab92d1fa --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d3_4673f2297c2b", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic shelf with a white mug in a row of similar mugs, clean boundaries and high contrast for easy separation", + "target_evidence": "white mug", + "expected_visual_elements": [ + "shelf", + "mugs", + "similar objects" + ], + "negative_prompt": "low resolution, overlapping elements, watermarks", + "diversity_tags": [ + "shape similarity", + "layout variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "4673f2297c2b", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic shelf with a", + "max_similarity_to_other_prompt": 0.6667, + "diversity_score": 0.1333, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_4673f2297c2b", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_4673f2297c2b/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29062757, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2220584, + "qwen_elapsed_seconds": 5.166, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..0e98d27b54f3544df773bc0675ad95a551de5104 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff3b67251ea1baee07075676ac9c2bae3d1951e0ad693ac9ed9109d58e89c4bd +size 545474 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..a81fe6a0a7e3b550a9dce9da80d216abd3e58111 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a519be6276079e307d2e38a36377462a67ae793134c05e665960f434749b4b31 +size 696653 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..5ebb6d8edb867b411d39488d6c8841943fe15456 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:66626a0430134507270d8b7781a2d3a1fc0903943967338a0f7042f59c3a02a4 +size 582218 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/record.json new file mode 100644 index 0000000000000000000000000000000000000000..1a8cbabc715dccbcb720f2fce4f53703427065a9 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d3_4f0d15b53f93", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic market display with a yellow calendar on a stack of similar colored items, sharp focus and clear separation", + "target_evidence": "yellow calendar", + "expected_visual_elements": [ + "market display", + "stack of items", + "similar colored objects" + ], + "negative_prompt": "blurred text, overlapping objects, watermarks", + "diversity_tags": [ + "color similarity", + "object scale" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "4f0d15b53f93", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.5, + "diversity_score": 0.3, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_4f0d15b53f93", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 22684801, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 4249307, + "qwen_elapsed_seconds": 5.171, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..1ddbd007160ff3a243cda87c03e8aa750a7e5454 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c61feca3f644b0742557e6400e045b713e11ff99d6cb0c679ea41f33769a7eca +size 85157 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..c275b923fbc6dbb8a4bc7b7fa62e044f27e3a80c --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dad173bc384f7d16d176ad695e54f1fd1071cfb30fa8c5ecd14e8465c3b01f35 +size 544526 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..9be1de04ba74d5e9a8b2af944bc2f4270cad5e55 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc2717ed6c20c0dd4e90d202ca9166b6222c1b2b82d1ccbfc22413e179ccf5b0 +size 603784 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/record.json new file mode 100644 index 0000000000000000000000000000000000000000..bc433a4eea27c94b4d28f26b7b0bf9179dab6af5 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d3_56877d3370c6", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic locker with a purple eraser among many erasers and stationery, high contrast and clear foreground focus", + "target_evidence": "purple eraser", + "expected_visual_elements": [ + "locker", + "erasers", + "stationery" + ], + "negative_prompt": "low detail, overlapping objects, watermarks", + "diversity_tags": [ + "color similarity", + "object placement" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "56877d3370c6", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic locker with a", + "max_similarity_to_other_prompt": 0.44, + "diversity_score": 0.36, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_56877d3370c6", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_56877d3370c6/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28241459, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2188272, + "qwen_elapsed_seconds": 5.17, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..a04e5eb13366a70cf17f0a5f77be4a2eaf4a5076 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b2e774ba226de40daede564f038882157f3f9b478b1644b1d93aeed44e9143b +size 405642 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..e08bc673ea1bc992142b58f0de54836a80205582 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eef5f79d9fd2428332f7473fb7fde853583d60e1a25c6f3b40443f1e9b92f68e +size 668339 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..fce4f69bb551923c0606e46d5d51726735af380b --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b9856293853cd821aa50128e733f48dca056d7466bb73f54a3f3ecba6bd44a0 +size 608393 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/record.json new file mode 100644 index 0000000000000000000000000000000000000000..3eed3ec84691b6ba5d55f6813562a31a8adf75e4 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "clutter_distractor_d3_596c2e1f65c8", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic workbench with a red ruler among rulers, measuring tapes, and other tools, clean boundaries and sharp focus", + "target_evidence": "red ruler", + "expected_visual_elements": [ + "workbench", + "rulers", + "measuring tapes", + "tools" + ], + "negative_prompt": "blurry focus, overlapping elements, watermarks", + "diversity_tags": [ + "shape similarity", + "object scale" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "596c2e1f65c8", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic workbench with a", + "max_similarity_to_other_prompt": 0.44, + "diversity_score": 0.36, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_596c2e1f65c8", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29983416, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1344084, + "qwen_elapsed_seconds": 5.119, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..32054297382ae3bf5cd75f322ec6592b72a9da64 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f3f3c7fc5e2416ccd5874ce0cc5e8c49f3b940cb535ee7ea1f675bb335f7e3f0 +size 753200 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..db408d8d24390bb499f3cfab7066330d0fe57e37 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5073260770b934bf8b4fed05205c722a323af7a32f763525b0d2f0639b9c400a +size 588331 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..4c7caa8481dd62eff2d34f981963de212b01c2e1 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cae568a351813ad88b03312b9e7ac45d658405284764c487f24cdb2e71cb397b +size 599535 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/record.json new file mode 100644 index 0000000000000000000000000000000000000000..93a073ad5b152c116e30dd64542deffeaa19251c --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d3_5c9d227b32b4", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic bulletin board with a blue notebook amidst similar-shaped paper items, sharp focus on notebook, high contrast", + "target_evidence": "blue notebook", + "expected_visual_elements": [ + "paper items", + "bulletin board" + ], + "negative_prompt": "blurry items, low contrast, overlapping objects, poor lighting", + "diversity_tags": [ + "similar-shape distractors", + "high contrast" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "5c9d227b32b4", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic bulletin board with", + "max_similarity_to_other_prompt": 0.44, + "diversity_score": 0.36, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_5c9d227b32b4", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24649423, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1950292, + "qwen_elapsed_seconds": 5.182, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..14a4aa21b05386092bfb76b3173d3b1d004bc3f6 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:88f6fa1c95a894707952fe82ba967ccc2545efbad0798a32963838a5998c7bcb +size 797609 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..ac72fccbae287d991cc3c1136ca3da34bbd39af8 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b44a3b2e5ed3772d875db0ab16eb6547226a3cc0f477508c44e1a8ab116753b5 +size 621872 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..beebe7e478ca3cf0641bca4aa1b529bfd9b1445e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7c4fa6ae61bf5066d296d5e3fb4826d959d202b6bbb21729403a3c22451a80d2 +size 615148 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/record.json new file mode 100644 index 0000000000000000000000000000000000000000..217b238533e86a1c8f00f765a98d964528a896c9 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "clutter_distractor_d3_8c9c64b73906", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic market display with a red traffic cone amidst colorful balloons, confetti, and paper streamers, clean boundaries for easy separation", + "target_evidence": "red traffic cone", + "expected_visual_elements": [ + "traffic cone", + "balloons", + "confetti", + "paper streamers" + ], + "negative_prompt": "blurry objects, overlapping elements, low contrast", + "diversity_tags": [ + "color similarity", + "layout variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "8c9c64b73906", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.3793, + "diversity_score": 0.4207, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_8c9c64b73906", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_8c9c64b73906/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28250007, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6699584, + "qwen_elapsed_seconds": 5.142, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..ddadb8b104f7d1fdde3ebf0c665b10b56f6fb4bd --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e8a932a64fc5a7eb6b356ba2132dd5fe2c2374fefecc9c6361a25751c6d26d4e +size 184990 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..f7141a81f37b1f116b1a1cf9f68c8015aa3e8251 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:92f82489d7a0b15901a99672ab64c29ad4bd0129309f972acd707ae75ad81173 +size 643832 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..885e0f378e6e83903eca072f7df9272eeb3d3bb0 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4df08e1bfce2d55dac74943fd7e49d473654244fea3b4ce564b0e1dc5ecc5d78 +size 621424 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/record.json new file mode 100644 index 0000000000000000000000000000000000000000..6edcd66714a5e9d3b34212a0a9cb5f23ec6ddb3a --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d3_b05af666789f", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic market display with a blue pen in a group of similar pens and paper, sharp focus and clear foreground separation", + "target_evidence": "blue pen", + "expected_visual_elements": [ + "market display", + "pens", + "paper" + ], + "negative_prompt": "blurred text, overlapping objects, watermarks", + "diversity_tags": [ + "color similarity", + "layout variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "b05af666789f", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.5, + "diversity_score": 0.3, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_b05af666789f", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_b05af666789f/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 21139338, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1276532, + "qwen_elapsed_seconds": 5.164, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..3aa06f4314d2f62808afd206876e5cdccb240791 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:29a51b76c27eb62dae9cc73e0a3bc714ad78d1205f529dc34cec8a90202b5c6f +size 520780 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..72907c7b4dcaca4b6bc749b6ad6cce08693ff2c0 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:616fb223087fb871c09cfce2e9d8f158e8476f5b7a5a9a820813ebb137dcebf4 +size 628922 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..e19a4ea6e2e19ff47922feb36db8ff60ed034035 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a5bb78ff3e9cdccbe6fe789e456fc5dfbbfe13cd5ee71f01ced126bdfce74e9f +size 606848 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/record.json new file mode 100644 index 0000000000000000000000000000000000000000..ca5c9da842f5743519489a3e4273df08b782d15a --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d3_b0fda5852ae2", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic market display with a red rubber duck surrounded by similar-colored plastic toys, sharp focus on duck, high contrast edges", + "target_evidence": "red rubber duck", + "expected_visual_elements": [ + "plastic toys", + "market display" + ], + "negative_prompt": "blurry background, low contrast, overlapping objects, poor lighting", + "diversity_tags": [ + "similar-color distractors", + "high contrast" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "b0fda5852ae2", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.5, + "diversity_score": 0.3, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_b0fda5852ae2", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29488989, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9192459, + "qwen_elapsed_seconds": 5.135, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..aa7142d3b9fde807e845da5508189ca19a5a2f19 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2bec2279f6164d2db78689902181bfbe2a3bc6b76ebb8bf2bd431d5791044ba6 +size 108317 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..ce3d37c6776862570a55ff035094d002b545c85c --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b87ad185c8ff9cef4c35146069945a4eaaa8397e604b4e15d483234054472922 +size 519871 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..d6bec0529d19c7f67eae5497145d5fbf292a6710 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b691393ece15f925c73e6da5146a9d09ef07461a1915bcdaca318e86a728357d +size 634604 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/record.json new file mode 100644 index 0000000000000000000000000000000000000000..8e4addffb8311d18da7a14fb3c1a9ef302890e7a --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d3_c3ccd586d6ae", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic shelf with a yellow book in a row of similar books, clean boundaries and high contrast for easy identification", + "target_evidence": "yellow book", + "expected_visual_elements": [ + "shelf", + "books", + "similar objects" + ], + "negative_prompt": "low resolution, overlapping elements, watermarks", + "diversity_tags": [ + "shape similarity", + "object scale" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "c3ccd586d6ae", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic shelf with a", + "max_similarity_to_other_prompt": 0.6667, + "diversity_score": 0.1333, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_c3ccd586d6ae", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 27839720, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5146656, + "qwen_elapsed_seconds": 5.101, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..b18b60be4752d657182328f0011688f7ee507cfa --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74f6073e0190ccefadce2538ee95da5e5244f8b9d3dcbfcaba7b66edcdb75df8 +size 408439 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..5b4b792bdb376455fe5fa7ac6f3b6d6a40e8f1af --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7423491c7b94f6ac72752bcd18eef62a76acf432c1a8728214f12939c343916f +size 778458 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..4c4c5beebad7245e43109a678b1b53b18bcf7f4e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d53fa05d798a7b1bb333d8a8c2394b2c67784f3f6f033ba6a0a0b20c748993f7 +size 629402 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/record.json new file mode 100644 index 0000000000000000000000000000000000000000..05820ac93bededdf7e92da3494e40c4377873268 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "clutter_distractor_d3_c685f2fb5896", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic bulletin board with a green pen on top of sticky notes, pens, and paper clips, high contrast and clean edges", + "target_evidence": "green pen", + "expected_visual_elements": [ + "bulletin board", + "sticky notes", + "pens", + "paper clips" + ], + "negative_prompt": "out of focus, overlapping elements, low detail", + "diversity_tags": [ + "color similarity", + "object placement" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "c685f2fb5896", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic bulletin board with", + "max_similarity_to_other_prompt": 0.6538, + "diversity_score": 0.1462, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_c685f2fb5896", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_c685f2fb5896/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26948933, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2772937, + "qwen_elapsed_seconds": 5.227, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..1a732a8f795b9dfb3bede524e60e80ff65035eac --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6258baa8d6a567ca5cd46ec8df36b8901c27951d65ebc525ef62ece92a15b2f3 +size 567016 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..5e7f27b54eff829f32e35bdfc76167c402657f5e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb65ccd2ed624ffd12a5eec2edc51ecc38f8cfdbc986339ce9e7d5d6b9ac5867 +size 585650 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..4c2d2661a8d4b579b0f5f7c71e8c95ccb53f2df2 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:06ee48ec57e30e0960751f914270c09f0db2821ebb7010bf3877c630b7664221 +size 570299 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/record.json new file mode 100644 index 0000000000000000000000000000000000000000..0a7af92506c3ed6ebb2198202d8c82d61d7685a5 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "clutter_distractor_d3_cbfa252c1b7d", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic tabletop with a white eraser among pencils, rulers, and paper clips, sharp focus on the eraser for clear identification", + "target_evidence": "white eraser", + "expected_visual_elements": [ + "eraser", + "pencils", + "rulers", + "paper clips" + ], + "negative_prompt": "low resolution, poor lighting, background clutter", + "diversity_tags": [ + "shape similarity", + "object scale" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "cbfa252c1b7d", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic tabletop with a", + "max_similarity_to_other_prompt": 0.4815, + "diversity_score": 0.3185, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_cbfa252c1b7d", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23354054, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7591050, + "qwen_elapsed_seconds": 5.203, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..3342f3dcfd9e265b8def3fec2f4a34cf3f3b1634 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d47a5eada7582f7d9e8abd6c592df2a7d9219d6629088c70db867847fa4af1aa +size 661712 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..ef26120b29be17bbbb4523c05ebafc1eed6af993 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7fd079f2e15dd4f5dfeb783128a61b6d1b6e0902157629662d541628b480a3bc +size 674122 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..9e995ccd5c92f6d860d89d6eff44ff34fb870b46 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e5f2ecf441326a58fcd2ed48a216406c7dcc79d50a9287061ae4c1eca0460800 +size 619240 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/record.json new file mode 100644 index 0000000000000000000000000000000000000000..934522b019c03511dc8d361e01caf07fce4e7968 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d3_d2419f42b8a8", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic market display with a green bottle among similar-shaped glass containers, sharp focus on bottle, high contrast", + "target_evidence": "green bottle", + "expected_visual_elements": [ + "glass containers", + "market display" + ], + "negative_prompt": "blurry items, low contrast, overlapping objects, poor lighting", + "diversity_tags": [ + "similar-shape distractors", + "high contrast" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "d2419f42b8a8", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.5417, + "diversity_score": 0.2583, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_d2419f42b8a8", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 21513032, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1374224, + "qwen_elapsed_seconds": 5.105, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..b78da4dc439081351749989f35ef4791bc7df63a --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d97d7bb762e4c783edbe4c70124a46283f6711361d6251c9d0871bd1b1d2b778 +size 558960 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..30a3ae5f77549ec6c5ff50fd0762d83353d305fd --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1d8d8aa0bf93b1d6fb54c42c8405d7cc76d209a8e99c07d9bd14a0acc4d8b8d +size 541975 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..6824669b0c5fadf59c926a4d374f338ef0077c5e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1b8c9451ab3a84eb0f98b6673c33c6033a71c7f411698aae8efe6680f1e452e5 +size 586253 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/record.json new file mode 100644 index 0000000000000000000000000000000000000000..c7206427bde31fe2896327a866c9a9689786622a --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "clutter_distractor_d3_eb19a4572a58", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 3, + "image_prompt": "synthetic photorealistic locker with a green ID badge placed between keys, coins, and pens, high contrast for visual separation", + "target_evidence": "green ID badge", + "expected_visual_elements": [ + "ID badge", + "keys", + "coins", + "pens" + ], + "negative_prompt": "faded colors, poor lighting, background clutter", + "diversity_tags": [ + "color similarity", + "text region" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "eb19a4572a58", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic locker with a", + "max_similarity_to_other_prompt": 0.3846, + "diversity_score": 0.4154, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d3_eb19a4572a58", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_eb19a4572a58/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25352516, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1503295, + "qwen_elapsed_seconds": 5.18, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..3a7cd6f814045327ba0a6e06f56eee702257dde1 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:66eadbb4b2fc6ec1aba23cba6de71be60fdf2ca51fcacf47533bd13b012d88b7 +size 400448 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..909890d18288fd79ad3f999ce84b01f2040a42c7 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:097ae62237b210aedaa6c81c8912158c2c395f62706778422c229dcbea7088a9 +size 733903 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..e2e5e68292127688b1ffa272e349ec0dc4eecb0a --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:47d9f10c8c83d4c1ccf14f6ee29a3e81b3279d4eea3a176b567297a1082f9515 +size 664630 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/record.json new file mode 100644 index 0000000000000000000000000000000000000000..d811e5bf223ee57a166eca2c8f2f9734d91f87c6 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d4_11a9bd230ad0", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic display with a black watch surrounded by similar-shaped gears and cogs, clear separation, sharp focus", + "target_evidence": "black watch", + "expected_visual_elements": [ + "black watch", + "gears", + "cogs" + ], + "negative_prompt": "overlapping items, low contrast, poor lighting", + "diversity_tags": [ + "shape similarity", + "object count" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "11a9bd230ad0", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic display with a", + "max_similarity_to_other_prompt": 0.44, + "diversity_score": 0.36, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_11a9bd230ad0", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24720661, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6061195, + "qwen_elapsed_seconds": 5.186, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..405d9caceb94092ef8ff0a78487e977c45a4d8fd --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9527f32945339c9555d76344cfeb9baf5cfbfe44fd38c84aa36d2382ba656085 +size 73872 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..09b2ecc9b01ebf775fb472264926ed63717d8839 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a5b8c28b9b962c72a59151b9e4586f8dd0d63127ea59fed91faf4b7dd64b4fcc +size 526202 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..226709d55c7d5df12e006f3ef409e834493c713b --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:747ba2e482fc001d46b251d0bc199bb46aceee361a5b900b2525d653fb5bb532 +size 510516 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/record.json new file mode 100644 index 0000000000000000000000000000000000000000..985b45d4f817d72f4e94d994cbe46558d747c968 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_23e2e0468879", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic locker with a single blue pen amidst many black pens, clean boundaries, sharp focus, high contrast", + "target_evidence": "single blue pen", + "expected_visual_elements": [ + "locker", + "many black pens" + ], + "negative_prompt": "low contrast, overlapping objects, blurry edges", + "diversity_tags": [ + "color similarity", + "object scale variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "23e2e0468879", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic locker with a", + "max_similarity_to_other_prompt": 0.5652, + "diversity_score": 0.2348, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_23e2e0468879", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_23e2e0468879/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25589732, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 607559, + "qwen_elapsed_seconds": 5.109, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..028d9c3d3ca78c07de69688eac6d5a24a85c1e16 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6670aeef05840f379b108c2b60c3ad1c3fecf5e45e3f7e09fc7522a834a5b8d +size 497280 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..5be39b78ada697dc9196ced2889c2d174e8ae1fb --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65ac574406b6aa7af2f726a6ccb5efbaad61ab1389e8a3dbf8677c7bd4c8b303 +size 628263 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..457a1fd36c0b39cf55d9b50fc2e6eb8f40f3ab45 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6820b83e70c4ca2714117369c25a0201dc894acbe049969db5610bcb7d2b7077 +size 574196 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/record.json new file mode 100644 index 0000000000000000000000000000000000000000..ca74575a90a4e4dcb744cbc93ab786d438add88a --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_263c71fe3951", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic workbench with a single green key surrounded by dozens of black keys, high detail, clear foreground separation", + "target_evidence": "single green key", + "expected_visual_elements": [ + "workbench", + "dozens of black keys" + ], + "negative_prompt": "low contrast, blurry edges, overlapping objects", + "diversity_tags": [ + "color similarity", + "object scale variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "263c71fe3951", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic workbench with a", + "max_similarity_to_other_prompt": 0.7273, + "diversity_score": 0.0727, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_263c71fe3951", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_263c71fe3951/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 27662153, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 741514, + "qwen_elapsed_seconds": 5.124, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..6c743b84188d93854d86ed9c96bfa98700501e91 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87f79f325352ce1a1e73dac7c7db0df6161249abb0e2edf2696d68de29ff9abf +size 690004 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..1699f252bef54ed93ad03a98e53fd3b499e46d7e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ffbabc2d6ebe610671d1ea2c3643fb4eeed8b528b4bb2dcd78531f1df316366f +size 776416 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..e7640907e0789d1d61343d6c796ee63e6ee3a1a0 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5722d49f7bfc3eb1d7688a7c17fdefa97cfc1086ab7357ffb235cb81b699562f +size 616277 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/record.json new file mode 100644 index 0000000000000000000000000000000000000000..d108c8b6afdf1c447caac0759eee7986449d0909 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_31b9fe844977", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic market display with a single red pen surrounded by dozens of similar pens, clean boundaries, sharp focus", + "target_evidence": "single red pen", + "expected_visual_elements": [ + "market display", + "dozens of similar pens" + ], + "negative_prompt": "blurry objects, overlapping elements, low contrast, poor lighting", + "diversity_tags": [ + "similar-color distractors", + "high density" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "31b9fe844977", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.5769, + "diversity_score": 0.2231, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_31b9fe844977", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_31b9fe844977/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 27294904, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7959812, + "qwen_elapsed_seconds": 5.235, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..f3a5239fa8acc8396d6d862efdf160745d90a1b3 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e27b3f31ae005b10842bda518bcdf53da1a40510b369d73fdd5ede5f1d1a0a6 +size 445861 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..51c7ff7fee9b1f9f6add08cf4171ed8eabd641a7 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0732b343b7268e7ab4070bbd51ad8d79e4385dae60be9df58cca15d27860efb7 +size 624166 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..7354c49f3f30cbe46f2da78dc9bb59ee8afb93c9 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fdca3d4c082534c5f284cf84bc7ce39f02b303e16dc54d2659c8ab81345527c6 +size 575505 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/record.json new file mode 100644 index 0000000000000000000000000000000000000000..0e944a1a1e38846299c71c944f9aa43ef66031e8 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d4_43a4872c9b38", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic market display with a single red apple surrounded by 25 similar-looking oranges, clean boundaries, sharp focus, and high contrast", + "target_evidence": "red apple", + "expected_visual_elements": [ + "red apple", + "oranges", + "market display" + ], + "negative_prompt": "blurry objects, overlapping elements, low contrast, poor lighting", + "diversity_tags": [ + "color similarity", + "high clutter" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "43a4872c9b38", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.6, + "diversity_score": 0.2, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_43a4872c9b38", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_43a4872c9b38/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25413465, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 8586496, + "qwen_elapsed_seconds": 5.103, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..c31b913b6be4512c9cd7e7f97b6e2ce2d0f97881 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d0dff19aef0c3d0745d2d9af366b9761c07704f3d8d5c085f58ca3ba1240469d +size 527697 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..07df747f534d76587f1b4150a0a25af08a1c7582 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0afa170c20a8f578053f9ca6f47a83d738dd59e035c1978e2c4301560736f25 +size 695832 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..bc8fec16ab14c26b3fa087116e3d4f43d7911185 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3261fb3f07b6d738b8935ddbbf46387d21beae040a186432574eaa5d8803760e +size 600995 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/record.json new file mode 100644 index 0000000000000000000000000000000000000000..090e6f75816964b18691bf782817cc8cd130ff0b --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d4_4444dce4f109", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic workbench with a single yellow key hidden among 35 similar keys, sharp focus, clear foreground, and high contrast", + "target_evidence": "yellow key", + "expected_visual_elements": [ + "yellow key", + "keys", + "workbench" + ], + "negative_prompt": "low resolution, blurred details, overlapping elements", + "diversity_tags": [ + "shape similarity", + "clean boundaries" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "4444dce4f109", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic workbench with a", + "max_similarity_to_other_prompt": 0.4444, + "diversity_score": 0.3556, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_4444dce4f109", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_4444dce4f109/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28370943, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 4187677, + "qwen_elapsed_seconds": 5.14, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..ab115c16102d2fb39d179de09abbf9e97d19b07c --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87e796572042672525f1a7dd29a1063659914ea158118e95407b4ba7e51c4521 +size 502660 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..a35fd6da3387586e2371d33beeb062ad83300292 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ade57e263e8e7c12a16ab29a43ece85806aad576e2038af96737bcc55302041c +size 568159 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..e2f968f835410da2b874e831c5c21f50d31c177b --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4644b258311fd2891b062ba719ee7900d94f800f3650708d453f7cdfea970c4 +size 604504 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/record.json new file mode 100644 index 0000000000000000000000000000000000000000..47dda9ddfd671d27e4ebcae31cba452238c8bc9c --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_454b748fb55b", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "Synthetic photorealistic market display with a single red apple amidst dozens of green apples, clean boundaries, sharp focus on target", + "target_evidence": "single red apple", + "expected_visual_elements": [ + "red apple", + "green apples" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", + "diversity_tags": [ + "color contrast", + "high density" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "454b748fb55b", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.6667, + "diversity_score": 0.1333, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_454b748fb55b", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_454b748fb55b/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25624714, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1774113, + "qwen_elapsed_seconds": 5.148, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..b084bf752b917579692e3cafc2de56353c2e1459 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:531f94ed99aae067141a5b2dd61c4456f2d10fc61140a96e87213ef6eb04f11b +size 324377 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..31824eeb149687dd0c22b18e8c551d28183e8bcd --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:227d406ebaf1ccb6eb0aa171d103c1e44417574596008402c98c93ef1e7d4be5 +size 580605 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..883421440d28b799ca827b5d33fd06406de232fc --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5d99fea521ba14a27edbc0168f83069be32d483e39cfaecda5e403b9342890cb +size 582773 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/record.json new file mode 100644 index 0000000000000000000000000000000000000000..cbfb6f0241ee37c4a885fe9e6b5579aa09910da8 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d4_5be1c3b7383a", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic locker filled with 30 identical blue pens, one unique green pen in focus, clean lines, and clear foreground separation", + "target_evidence": "green pen", + "expected_visual_elements": [ + "blue pens", + "green pen", + "locker" + ], + "negative_prompt": "distracting textures, overlapping items, soft focus", + "diversity_tags": [ + "shape similarity", + "object scale variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "5be1c3b7383a", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic locker filled with", + "max_similarity_to_other_prompt": 0.4138, + "diversity_score": 0.3862, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_5be1c3b7383a", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24595388, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6664834, + "qwen_elapsed_seconds": 5.217, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..0a96e72f1f3f3d9cf5d080c93f00bc0561f370d9 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0cabdc79119490489d8d6c4a2b1eb01af738524e80a212d3f03f20e1e2a9063e +size 371620 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..810d78d8b522cda7727d2c192ffc909dc5d59eba --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f281fa48d8c1357eb61967c4e20535001231e150e1c65bb1e4b84613a34f586e +size 412245 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..13bd891247cafb1795a6f59df926ab4d60f739a9 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ffdf7991da78e89519c0abe1386d408d233fff50bc9fe4fe5ceeee06e72e313b +size 586393 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/record.json new file mode 100644 index 0000000000000000000000000000000000000000..f503315c51eacfe472603512a6fed94d8dbd582c --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_5bfbc80b56a2", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "Synthetic photorealistic shelf with a yellow label on top of many blue labels, distinct color, clear foreground separation", + "target_evidence": "yellow label", + "expected_visual_elements": [ + "yellow label", + "blue labels" + ], + "negative_prompt": "blurry, overlapping, low contrast", + "diversity_tags": [ + "color contrast", + "placement emphasis" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "5bfbc80b56a2", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic shelf with a", + "max_similarity_to_other_prompt": 0.4583, + "diversity_score": 0.3417, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_5bfbc80b56a2", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26021712, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7542871, + "qwen_elapsed_seconds": 5.296, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..56bd0873f46a86db4ae2e6fd9ad4cdb693b9abea --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:040fb35880f8b6991dc6847c3e25941c739718fcf8dc90d6367670a1ba3662c3 +size 461179 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..9a2829d8572da3d52a21884fe89c71bac5ffd638 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:abf44133b0173e13840349a7f0acfb910612b7e0259c3f07ded9ff64fe22ac1e +size 560439 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..191291dc017e66cbfd2205b5e751b31463474b0b --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f85ea1cc96c4fc38f0486a7d77de21c4c31ec5b8d8e1bcff8c20968d712024b2 +size 539850 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/record.json new file mode 100644 index 0000000000000000000000000000000000000000..be9ed692532c4d6bd358f4816f164fe54c80a1b1 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_6c208faaaafa", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic shelf with a purple pen in a stack of similar-colored markers, clean boundaries, clear foreground", + "target_evidence": "purple pen", + "expected_visual_elements": [ + "purple pen", + "similar-colored markers" + ], + "negative_prompt": "overlapping items, low contrast, poor lighting", + "diversity_tags": [ + "color similarity", + "object count" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "6c208faaaafa", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic shelf with a", + "max_similarity_to_other_prompt": 0.4444, + "diversity_score": 0.3556, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_6c208faaaafa", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_6c208faaaafa/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26661592, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9993344, + "qwen_elapsed_seconds": 5.142, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..9c789fdbb026b8fd8087ac2cd5f048e9ec164af8 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ec7f8dced4969ba822bf54df2a35415e6b7ca0a948096d12bab72f470fee496 +size 610322 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..4de011a32da5f5fbcdb2e2a7582b68646dbf4105 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9cd7b526e7e14f98e2003512d6cc961c45e5b3d5781b79105de931b7f646b8b9 +size 609115 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..013e5b5092620fe9abbd3de2abb86d27729018be --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df0196195b76f4908c682db41debdd85da8c228cbad41337af50856a52c12bbe +size 592923 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/record.json new file mode 100644 index 0000000000000000000000000000000000000000..aff3f65046a33871f9ba9fbde44d77bd1b0e3ab8 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_71d67fddf097", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic market display with a single red balloon among hundreds of blue balloons, sharp focus, high contrast", + "target_evidence": "single red balloon", + "expected_visual_elements": [ + "market display", + "hundreds of blue balloons" + ], + "negative_prompt": "low contrast, overlapping objects, blurry edges", + "diversity_tags": [ + "color similarity", + "high distractor count" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "71d67fddf097", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.6522, + "diversity_score": 0.1478, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_71d67fddf097", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_71d67fddf097/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24219131, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 8726131, + "qwen_elapsed_seconds": 5.201, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..b69d99202149d9d3561b570287968428232701b8 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:505d6694395909aafd5f80e8ba495e9053ec8357734d0eea415d600340ec096a +size 629719 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..f97fa0c6a684d0471310ccf897168ac779ca03ce --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e2fdcd7833f056e48d008a3a1fd06012a5ce1a6bcab2f911fa734f64ac8fcd05 +size 535863 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..f2eced1e47cf856992316e69f3600f184fbf5ae4 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:04de205793ac67e3dda41fd655b8ebcdd3f76caf491b9e8a749243ff59123179 +size 665954 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/record.json new file mode 100644 index 0000000000000000000000000000000000000000..414f65585815c46a2d2228198fb4cfdcfd5ae629 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_7a819468a683", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic shelf with a single yellow key surrounded by dozens of identical keys, clear foreground separation, high detail", + "target_evidence": "single yellow key", + "expected_visual_elements": [ + "shelf", + "dozens of identical keys" + ], + "negative_prompt": "low resolution, overlapping items, low detail", + "diversity_tags": [ + "shape similarity", + "object scale variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "7a819468a683", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic shelf with a", + "max_similarity_to_other_prompt": 0.7273, + "diversity_score": 0.0727, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_7a819468a683", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_7a819468a683/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25753108, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6946850, + "qwen_elapsed_seconds": 5.233, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..dcd58cae7d7773a94b21eb01ea324f0e907efebc --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:80eb55538bc1bcfcfdb65826f009fd8860d66b9e9b4739964c2df8c579ee4b85 +size 404633 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..8229480cd3fb33e00549ea90e1bddd589f7d2622 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f381413074ed8897266ca5c3e6d5cabece0c041753b53120def28964e2c3216 +size 559936 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..f3afd18786a87f83edb151f798a8b74c77effa5f --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f1f6d6c4fe5e57e438063c5706c200b032def9f589b08cd42c6fc0e103edc3cd +size 696041 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/record.json new file mode 100644 index 0000000000000000000000000000000000000000..a93e22c7de6777585c38920c1dcf1df14baee3fb --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_88f59b021a79", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic workbench with a single white ruler among dozens of wooden rulers, high detail, clear foreground separation", + "target_evidence": "single white ruler", + "expected_visual_elements": [ + "workbench", + "dozens of wooden rulers" + ], + "negative_prompt": "low detail, overlapping items, blurry edges", + "diversity_tags": [ + "shape similarity", + "object scale variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "88f59b021a79", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic workbench with a", + "max_similarity_to_other_prompt": 0.5417, + "diversity_score": 0.2583, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_88f59b021a79", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_88f59b021a79/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26649291, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1063384, + "qwen_elapsed_seconds": 5.201, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..407bf9425c6b8554e4b42301794f793394f8e2a3 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb462aa69fc2e224b973fa3f40aa3e1c67d082c95fc38ccecaae2378d6195b9b +size 435879 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..4c10df41943f450a4c5b2a424fc942b0b5b92776 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83e146571aecf1beb71784625f1a297a361798482402228656629c7356f5cd18 +size 589929 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..81b4f67b733d112250d454b8052b726eb471bc7f --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c46144e041d0f9a9d1418f6225222500d50b810288f98d9eab80e9e12086c08 +size 571743 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/record.json new file mode 100644 index 0000000000000000000000000000000000000000..6647973ebf18956a00f9cff83e22bd083d7b8ebb --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_97fe5dcd2ae5", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic market display with a single white apple in a sea of red and green fruits, clean boundaries, clear focus", + "target_evidence": "white apple", + "expected_visual_elements": [ + "market display", + "fruits" + ], + "negative_prompt": "low contrast, blurred edges, overlapping elements, poor lighting", + "diversity_tags": [ + "similar-color", + "clear-boundaries" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "97fe5dcd2ae5", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.5385, + "diversity_score": 0.2615, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_97fe5dcd2ae5", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23011897, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 4374182, + "qwen_elapsed_seconds": 5.247, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..3ee6843dc3a4c2ebc0b4840012a23493b455a764 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5091b4d55f16ebef49f2ae362bbb68069690a69569cf46f2cd3522a860d561fb +size 525253 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..006e74ffe7aafa0c8dfce4ef7a80994b1c066fc5 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:019f9f9fe61bb88c112127780ee71ccb46d2516ed850316ae1bd074ca9a59bf2 +size 605156 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..c219657e3c62d104ebc666349acf2d3423eec24f --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2fdc8fcf2b707959e1a32c192fa420b5a24f644027a412d89cd3d02b3516de7d +size 584220 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/record.json new file mode 100644 index 0000000000000000000000000000000000000000..79e71b0ee75c062fb81e5e5c21dc2b8610c1687e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_9b4e5450a8e1", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic market display with a single red apple among hundreds of green apples, clean boundaries, sharp focus, high contrast", + "target_evidence": "single red apple", + "expected_visual_elements": [ + "market display", + "hundreds of green apples" + ], + "negative_prompt": "low contrast, blurry edges, overlapping objects", + "diversity_tags": [ + "color similarity", + "high distractor count" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "9b4e5450a8e1", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.7273, + "diversity_score": 0.0727, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_9b4e5450a8e1", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24204502, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 8857956, + "qwen_elapsed_seconds": 5.178, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..e0031037625ef51e79abbe55127f2b515eea97e9 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cae18ef22e3c2bcf782067fad309f0383b5e0f9bf7f32b0f1d6c83237999d6b3 +size 315782 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..a30c1ccce57a8a08980a4f18e0abd4e89f92a43b --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b3ed1befa14ef68ad9a4281ff0f4a5d34bacbbebf118aa07fa12b0cdac5878dc +size 781211 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..76151eab8f5b00a17088a932360d08c1258865e3 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:217beb00fa1945ae371bbc4e8d4980256353baaf6da5a2df08046faf4ad7eabe +size 532016 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/record.json new file mode 100644 index 0000000000000000000000000000000000000000..02f4fc2cc4eeb2c0218db0e2eaee515b0d82cbc2 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_a7a966f3fcee", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic bulletin board with a single blue sticker among many similar stickers, clear foreground separation, high contrast", + "target_evidence": "blue sticker", + "expected_visual_elements": [ + "bulletin board", + "stickers" + ], + "negative_prompt": "low resolution, out-of-focus, overlapping elements, poor lighting", + "diversity_tags": [ + "similar-shape", + "clear-boundaries" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "a7a966f3fcee", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic bulletin board with", + "max_similarity_to_other_prompt": 0.48, + "diversity_score": 0.32, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_a7a966f3fcee", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20326051, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6521731, + "qwen_elapsed_seconds": 5.223, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..1b4b0b1430183ff449b96bb90b22dc8c4242d926 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe8344730ed3790b75b00f68b0ff7bdf26caf26ce82a3caa8f98c4d0c56f096e +size 1061087 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..462a3a1fec1c599f97a2bf617c8f75dd8ad46d1e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4bac6a03c8d78e1c037aef8f8502fbe698826fbd4f11b81427988f1f25979526 +size 550083 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..679abefba3b7f2ec3e46ac1c985be3aba73f3332 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c964977b14adeffc79cd6352a0ff3de33620af6f6895a0650ce245439fe743f3 +size 570120 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/record.json new file mode 100644 index 0000000000000000000000000000000000000000..ce3b9c2b4f666fa3c9a3161d574630ce45a90535 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d4_c6f250be29ba", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic bulletin board with a green folder surrounded by sticky notes and rulers, high contrast, clear foreground separation", + "target_evidence": "green folder", + "expected_visual_elements": [ + "green folder", + "sticky notes", + "rulers" + ], + "negative_prompt": "faded colors, poor lighting, merged objects", + "diversity_tags": [ + "shape similarity", + "layout variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "c6f250be29ba", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic bulletin board with", + "max_similarity_to_other_prompt": 0.4643, + "diversity_score": 0.3357, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_c6f250be29ba", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_c6f250be29ba/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24048938, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2086878, + "qwen_elapsed_seconds": 5.207, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..26b7f796e5e772c5367e12d357796f0f55f7de3b --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca8f47ac1924f35813fbd85105ba71a8bc32c4d6d14faf3bc3b169f5e83edd1e +size 318185 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..bcb6703bb873dca6ba5ef78cf5a3ee5dd750f395 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b5b75a3d368e9e7ac4e65e69fb7ffaca6e65715bcaddae7b0c1b0dd7b41f324 +size 949776 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..cfadc3c330eb73647a03d0e2e34703402ac03278 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b26fd4e4e7afee894c51966f3866fd776e2f906daf7fb0c6182d03099d01fc48 +size 596193 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/record.json new file mode 100644 index 0000000000000000000000000000000000000000..03e6320b93921e5d14700fb72aba0066245cb6b0 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_ce19ddf16fef", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic bulletin board with a single green pencil surrounded by dozens of black pencils, high contrast, sharp focus", + "target_evidence": "single green pencil", + "expected_visual_elements": [ + "bulletin board", + "dozens of black pencils" + ], + "negative_prompt": "low contrast, overlapping objects, low detail", + "diversity_tags": [ + "color similarity", + "object scale variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "ce19ddf16fef", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic bulletin board with", + "max_similarity_to_other_prompt": 0.4615, + "diversity_score": 0.3385, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_ce19ddf16fef", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29519981, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2624438, + "qwen_elapsed_seconds": 5.184, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..1551a2306ae5ee85f94fe4ace80f190cd00e08c0 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dcd97a4f687bb9d65f0fbcd0881cd4e2d459027d025cc48ce0c2941ecc902839 +size 570868 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..1c15d448a08edcc57b4bb5194a2b0e70f89dd95c --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:64af2abf057bab353b770c9d4bf6589d0c0176b8765f0f69fa170265850d9278 +size 758911 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..a4830b5f6313a13bd89e4897fc2f4d1f41a62748 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e3bd18da610b6ee4411544972896707026f5c2c4622d315d8dd9a0b44b7e6f5a +size 606379 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/record.json new file mode 100644 index 0000000000000000000000000000000000000000..358e819dd16571b6117f4066204ca3df28ea7b47 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d4_ea4e61598d94", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "Synthetic photorealistic bulletin board with a white label on top of many colored labels, sharp focus, distinct placement", + "target_evidence": "white label", + "expected_visual_elements": [ + "white label", + "colored labels" + ], + "negative_prompt": "overlapping, low resolution, poor lighting", + "diversity_tags": [ + "color contrast", + "placement emphasis" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "ea4e61598d94", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic bulletin board with", + "max_similarity_to_other_prompt": 0.44, + "diversity_score": 0.36, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_ea4e61598d94", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_ea4e61598d94/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 22339680, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1371864, + "qwen_elapsed_seconds": 5.222, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..4be57aa98c242470088c0c81642cbfc7a1bd108b --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab4989142bf863b81138f949bb6a7b0b8e0d8bb23766076ea036f8748701823e +size 222416 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..bd67ce1ed97e963220b499210438cfb61ce8aa4a --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4836db51b3905f21ba6e1f06d590b5a9ea47ab119bc05a97cc21f04d3d9a6fed +size 739960 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..29f0a3a9f4368e1b729bae68cb16f43eb17efa44 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dbf9755254e74e77b7c327191f39a810ff4e6a956f12a5f4291e5e12279121b9 +size 587644 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/record.json new file mode 100644 index 0000000000000000000000000000000000000000..e8c95e3138c8998249d499e2df0aafbe90a6d7b2 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d4_ff2b8c5d1325", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 4, + "image_prompt": "synthetic photorealistic workbench with a blue key in a pile of screws and nails, distinct edges, clean boundaries", + "target_evidence": "blue key", + "expected_visual_elements": [ + "blue key", + "screws", + "nails" + ], + "negative_prompt": "overlapping items, low resolution, poor focus", + "diversity_tags": [ + "object scale", + "material variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "ff2b8c5d1325", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic workbench with a", + "max_similarity_to_other_prompt": 0.3704, + "diversity_score": 0.4296, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d4_ff2b8c5d1325", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23168712, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9683140, + "qwen_elapsed_seconds": 5.24, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..c9c01897e7f533ba204869c50aa61d75e397bf5b --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ce8fe251f507bcf4672e611f90a8b3d381f624247e049b6c68109e53c93daa8 +size 369133 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..e266795517e7ba61fa78c95e31b0fb719913ccb8 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77af2617917239b348038a1a92a0e5bd04aa0f21ce92f010a74ad5e6dc906907 +size 523261 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..d0b60a63daf7ea6b67b8275ca230177642cfc705 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:34f52ef46f512a1a3fa718a02c998fab1ef0c909db1d0a9ca77ba141e12caa2e +size 747203 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/record.json new file mode 100644 index 0000000000000000000000000000000000000000..51c03ad85da68ad42acbcbba22401b37bf39b9e9 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d5_04829301b77b", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "synthetic photorealistic market display with a red wrench among scattered tools, sharp focus on target, clean boundaries, high contrast lighting", + "target_evidence": "red wrench", + "expected_visual_elements": [ + "red wrench", + "scattered tools", + "market display" + ], + "negative_prompt": "blurry objects, low contrast, overlapping elements, poor lighting", + "diversity_tags": [ + "color similarity", + "tool clutter" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "04829301b77b", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.5833, + "diversity_score": 0.2167, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_04829301b77b", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_04829301b77b/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25227361, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 4186258, + "qwen_elapsed_seconds": 5.215, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..765581bb38b4f8555682efb5b75b11ba0f2702fc --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:42c5bf9ca552ef2709712bc6a6e88c9c43810e2c6bb9f2c473d1c59c5cd9c87b +size 307541 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..52aeb10fc13f6bb1b30349641c53cc4e000524bc --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e5d1c3009fc7ecaf7f7348893ba10456aa969d9b1d12a9ea8b5f1d59e56516c +size 504176 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..e5328bd044add51d250c265fd449065db465b570 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:615414999dfde65fa2237d541e89fba39410e10ee6431a7d2dee085e6d93e070 +size 580501 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/record.json new file mode 100644 index 0000000000000000000000000000000000000000..a7bd146f5c5844f1c89d21d7c5adc4180b390b74 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_06238c20c828", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic tabletop with a single white cup among 40 similar-shaped ceramic objects, high contrast and sharp focus", + "target_evidence": "white cup", + "expected_visual_elements": [ + "ceramic objects", + "white cup" + ], + "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", + "diversity_tags": [ + "similar-shape distractors", + "sharp focus" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "06238c20c828", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic tabletop with a", + "max_similarity_to_other_prompt": 0.4615, + "diversity_score": 0.3385, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_06238c20c828", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_06238c20c828/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29244267, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9106413, + "qwen_elapsed_seconds": 5.22, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..6ab19f4d597c5c4f0b0e052107e4f114cfb3b665 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f8eed00c22d0d0a6bd22a9530207c84c8a36bb3ad3b461aa9cccea17f30a2dcd +size 323623 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..f8ab9750d8b89e81005c6d00a9f0683fe01483e6 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:608bbe0c6a0824b977e76f712820a1505c854581fe588cc783cc8d41acb44452 +size 557070 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..beee3224f5898df996139c55fd1dbca666409885 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e99ff22ec898d3961ac6a555a4bb937f48814fff0353bd9f4c754ffc8f9524e6 +size 522687 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/record.json new file mode 100644 index 0000000000000000000000000000000000000000..0d40399667d3491b558b6e405d50ff8f7a54b2b4 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d5_1aa08434a01a", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "synthetic photorealistic shelf with a red pen surrounded by similar-shaped pens, sharp focus on target, clear foreground", + "target_evidence": "red pen", + "expected_visual_elements": [ + "red pen", + "similar-shaped pens", + "shelf" + ], + "negative_prompt": "overlapping pens, low contrast, poor focus, blurry background", + "diversity_tags": [ + "shape similarity", + "office items" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "1aa08434a01a", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic shelf with a", + "max_similarity_to_other_prompt": 0.6087, + "diversity_score": 0.1913, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_1aa08434a01a", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_1aa08434a01a/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26731752, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1239610, + "qwen_elapsed_seconds": 5.233, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..8e45ea08928b017fa66c9313f3b1a6480a6664a6 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c0167f2f6d8bf219cf80a17476ed21572020e0729e42eccc75dde32f8e20151 +size 565247 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..12494f36ded89c45b6b187f2be2bcf44f3d89e89 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c187b90a3cfbe3725de1c4781caaa4b542ce386d2566ff84761c60660eea9d0 +size 726333 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..9f7ed72fa5d8995bcdb78ce1c6aef1a0fa3d0608 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3347553cbf4cb5f403b819f4ed5ac68bf2746e4dd208f6cc45bff064e39da994 +size 560133 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/record.json new file mode 100644 index 0000000000000000000000000000000000000000..974ae2a0a7300903172a66ec2025c50ca2a4fe50 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_24ee9c8ef6d0", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "synthetic photorealistic market display with a single red keychain among 45 plastic bottle caps, clean boundaries, sharp focus, high contrast", + "target_evidence": "red keychain", + "expected_visual_elements": [ + "plastic bottle caps", + "red keychain" + ], + "negative_prompt": "low contrast, blurry edges, overlapping objects, poor lighting", + "diversity_tags": [ + "similar-color distractors", + "high distractor count" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "24ee9c8ef6d0", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.6, + "diversity_score": 0.2, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_24ee9c8ef6d0", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20598557, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7799995, + "qwen_elapsed_seconds": 5.222, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..eabf9fee3aae54ac27a8f5e71488ba8998bcd3db --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1d98b34033fd99c58c27e1a49852b666601f8ade13a76090c5d2430354e7368 +size 650840 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..d44e20a1da0fde1bc6a570a786a47eb2f0f8f15b --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3a9380d1c05efdb95f24828b64e7d26f79e1e2bfcb229cf095efd14c646bd475 +size 480039 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..e7ffecfd66ca01024661cdf200312d56f3b45ab7 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f64efb4c97282373a70e586b81aca558c33135dc51cbcb3a8100e19d25d9d71 +size 519334 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/record.json new file mode 100644 index 0000000000000000000000000000000000000000..3da69380a7033bdc69d8c5d5cba9be977fb6a9e9 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d5_2e2224605b5c", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "synthetic photorealistic bulletin board with a yellow label amidst similar-color labels, clean foreground, sharp focus on target", + "target_evidence": "yellow label", + "expected_visual_elements": [ + "yellow label", + "similar-color labels", + "bulletin board" + ], + "negative_prompt": "overlapping labels, low contrast, poor focus, blurry background", + "diversity_tags": [ + "color similarity", + "label clutter" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "2e2224605b5c", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic bulletin board with", + "max_similarity_to_other_prompt": 0.44, + "diversity_score": 0.36, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_2e2224605b5c", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_2e2224605b5c/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29785390, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1684863, + "qwen_elapsed_seconds": 5.168, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..71d8d6b656b1071c37f24235d4a7a9c5bcbadc63 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:746d1384686b49cd6e34a7dc32008d06a2818e7b0e8b60f74bcaa2987035a9ab +size 346319 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..63e426d81e061ca7bfb497821e289fdd83049833 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86424e182194d6d1a0f1412ea4346a96f972f791291a23ccd8d947e3501c5e4c +size 799123 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..5d19dfd2a129fe72a4e4189147670a0e8bc761ec --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:39f50c114daedda60dd962cbce8ddb27fa4549a1c7d700624a7e6394dfe08d18 +size 617675 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/record.json new file mode 100644 index 0000000000000000000000000000000000000000..c6752f2730339372317d1f96a9806c9ed79fe2c4 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_3979ce7f86ea", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "synthetic photorealistic shelf with a single blue USB drive among 50 white paper clips, high contrast, clean edges, visible target", + "target_evidence": "blue USB drive", + "expected_visual_elements": [ + "white paper clips", + "blue USB drive" + ], + "negative_prompt": "low contrast, overlapping items, poor lighting, low detail", + "diversity_tags": [ + "similar-color distractors", + "high distractor count" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "3979ce7f86ea", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic shelf with a", + "max_similarity_to_other_prompt": 0.3333, + "diversity_score": 0.4667, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_3979ce7f86ea", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29446383, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7962205, + "qwen_elapsed_seconds": 5.256, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..01bf57011cdd5ebf2140d43a58c4a2afe5b02535 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:995d72aa5e7393601dc9f8085be6f5ec2104ba0ed8a86605b2d624c36e839c2f +size 469330 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..59712335524c15da282be54788664da77505dd38 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e64038aacff0254546be8cf94b668851203e0f8a8d949571e62c58b15f1165f +size 583934 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..a86d14e91d334c6a4dae81b7c7aa8137591114b0 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:115471f3901a6a61fdaf4e0e678fa9f4401c106656c1f0ab657b0b6fb65019fb +size 566797 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/record.json new file mode 100644 index 0000000000000000000000000000000000000000..0974089752d770995258c5151f4ab9121f28132e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_62c3301efb9c", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic display with a single orange box visible among 40 similar-sized cardboard boxes, clean boundaries and high contrast", + "target_evidence": "orange box", + "expected_visual_elements": [ + "cardboard boxes", + "orange box" + ], + "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", + "diversity_tags": [ + "similar-scale distractors", + "high contrast" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "62c3301efb9c", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic display with a", + "max_similarity_to_other_prompt": 0.5, + "diversity_score": 0.3, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_62c3301efb9c", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_62c3301efb9c/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 22543368, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 8675430, + "qwen_elapsed_seconds": 5.192, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..059a6ff6c572a39a8e39fb4cbd45f0b4e08519e2 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9dfcea89990973699cef6571dfe46cad801d0737bea2d17572c379c70c0b112d +size 583568 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..7675c433e72752787a6c04dabb5298ba7430fc24 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2c99e055c065f72a4b85c826c5380a96bde05bf32dea8902aedd745d9903abb2 +size 641588 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..72255422a170bf21ed3fbfe940fd76ff67acdbc6 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f66ed3565f4a5c58da6a28e87a77f3051d935e526e6c6c9a7c271634321a6e56 +size 614802 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/record.json new file mode 100644 index 0000000000000000000000000000000000000000..9949f3bb632492544bc6aa67c50cbe873fdee843 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_637cd118a043", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "synthetic photorealistic workbench with a single yellow padlock among 38 metal screws, clear foreground separation, sharp focus, distinct textures", + "target_evidence": "yellow padlock", + "expected_visual_elements": [ + "metal screws", + "yellow padlock" + ], + "negative_prompt": "low resolution, blurred focus, cluttered background, overlapping objects", + "diversity_tags": [ + "similar-shape distractors", + "distinct textures" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "637cd118a043", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic workbench with a", + "max_similarity_to_other_prompt": 0.4444, + "diversity_score": 0.3556, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_637cd118a043", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_637cd118a043/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26491783, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7792357, + "qwen_elapsed_seconds": 5.175, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..0bf404e6cc3870cb3a73985346d8401ce3eb6bc2 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3d94f5a72ab05e8d51cbfc187ea17bef2443bc2a1c8570635657c90e67176f9c +size 390312 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..eb756ea667f6094e6f3edec6d8a270a5eb0f4f73 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:edf526e63fd0fcbdb88c589b549a2bd094a6603f703dd2876c99897b5d189fdf +size 643501 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..13005af0675cb8e8d85b38ec5403de1136e62414 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d50680aad341018b426e5e0ddf10c041081d581d9c0c01bd6c2b164fe97388f +size 577510 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/record.json new file mode 100644 index 0000000000000000000000000000000000000000..65b58625f6fb48dffdc3a9a1839d5af3f4b51c03 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_79441f6c0574", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic workbench with a single green bottle among 40 similar-shaped glass containers, high contrast and clean boundaries", + "target_evidence": "green bottle", + "expected_visual_elements": [ + "glass containers", + "green bottle" + ], + "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", + "diversity_tags": [ + "similar-shape distractors", + "high contrast" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "79441f6c0574", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic workbench with a", + "max_similarity_to_other_prompt": 0.56, + "diversity_score": 0.24, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_79441f6c0574", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_79441f6c0574/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26333002, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6156372, + "qwen_elapsed_seconds": 5.158, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..4de444bda4013515ae32619a5ef3cfca6bdef5db --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b9299c64e25b0ba29e8b0b4a2de8e2a57ab26b51b766c8b81c8c2daa77df061 +size 591441 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..2a289a32c178af20a1ed2f41e87d42e730238554 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ead84182ae04b9ffe9eddbfafa3574f32544aae1e2e75cbdf7b11c4017edb4c3 +size 506551 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..2a635f301c35ee632d6e5078ae180ccab0a74ae2 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b5d5ac5de604047776347e959f3ed49c85cf919c96c05f36d267ef83b80f8fe7 +size 703190 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/record.json new file mode 100644 index 0000000000000000000000000000000000000000..6a314800f7b2a527215b9ceaeb5cdf8b51a1020e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_82234aef50b3", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic bulletin board with a blue label standing out among 40 similar-colored paper notes, clear foreground separation", + "target_evidence": "blue label", + "expected_visual_elements": [ + "paper notes", + "blue label" + ], + "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", + "diversity_tags": [ + "similar-color distractors", + "clear foreground" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "82234aef50b3", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic bulletin board with", + "max_similarity_to_other_prompt": 0.48, + "diversity_score": 0.32, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_82234aef50b3", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_82234aef50b3/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28422949, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5478104, + "qwen_elapsed_seconds": 5.195, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..ff23aaccea2798a113e829495a237e5584112f17 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:27cc589168e0e19c7e1e0682917b0037f33c168fa002fbca5c6f98b52ce22645 +size 419107 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..a73cc1aa48d477bd31869dc7842f996cf35614b3 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:73d72674dd7a6736c825d2650729adc9df019f71bab88e76b657ac19f8ccc20f +size 562115 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..03cab026346884e5f0ac4827db45e6b8eb236388 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:93a3e8a9522107822f31b36b539d59b4e28da2868d372bea5e0adfaa26697d4d +size 677356 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/record.json new file mode 100644 index 0000000000000000000000000000000000000000..b8051cd7f59ccbd079cc01a767efbdb40d42dc8e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_85f68bfdbfa5", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "synthetic photorealistic locker filled with a single orange ruler among 37 small tools, clear foreground separation, high contrast, visible target", + "target_evidence": "orange ruler", + "expected_visual_elements": [ + "small tools", + "orange ruler" + ], + "negative_prompt": "low contrast, blurry edges, overlapping items, poor lighting", + "diversity_tags": [ + "similar-shape distractors", + "high distractor count" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "85f68bfdbfa5", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic locker filled with", + "max_similarity_to_other_prompt": 0.4074, + "diversity_score": 0.3926, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_85f68bfdbfa5", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23085812, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6095068, + "qwen_elapsed_seconds": 5.139, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..a1e6b9dcbc19278af22c7f6d140907f9c0df0044 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:438a55862e64d3bdce17b8174ebcdc15050872cddcf460d8fa82b249438e7c28 +size 557078 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..b36b9d9e019418ff686a9cf9e7f55cbe6361ae44 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:10f83787b85367df8c84e58279eb91a6694f07f60d0500380e57d0a3ecfc02f2 +size 564793 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..2ce201e7ed23aaaedd4d46e8c46be094bdb60c95 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b185867f418f257e160f2977c0e8d83c135034a97f95138aeaf75458b804a205 +size 571027 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/record.json new file mode 100644 index 0000000000000000000000000000000000000000..51b2dbda8916122668332dbe8ea3eea57694a270 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_9bc239a79071", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "synthetic photorealistic locker filled with multiple small tools, a yellow wrench stands out with clean edges and sharp focus", + "target_evidence": "yellow wrench", + "expected_visual_elements": [ + "tools", + "locker" + ], + "negative_prompt": "low detail, blurred edges, overlapping objects", + "diversity_tags": [ + "scale variation", + "color similarity" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "9bc239a79071", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic locker filled with", + "max_similarity_to_other_prompt": 0.5417, + "diversity_score": 0.2583, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_9bc239a79071", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_9bc239a79071/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23987141, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2830834, + "qwen_elapsed_seconds": 5.138, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..9d0fae47ccf072e94702f4d27a4aa8a72c074331 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eb8517bdfd1909c640d3ce091fffa48f12fac45e38f3af64a9c4310808f7f3aa +size 849937 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..2e0d1f9c4b0a4ce499829d7db212278d75dd7bb5 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:37ac0c86a2856c1b96564029a7e95050311f11f77f589c8279b11f5f73986743 +size 583965 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..7f819adb1e0784887ff4a4f49bb867e0a969e0e3 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2eaaf092f96248d35ae84d2468816dbeb0a2c1a743854aea7be9fc51ca44e55 +size 590301 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/record.json new file mode 100644 index 0000000000000000000000000000000000000000..41c18bcdda2300c1a26741d253cc4a218cff9cab --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_a3a3c294f30d", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "synthetic photorealistic bulletin board with a single green notebook among 42 sticky notes, sharp focus, clean boundaries, distinct colors", + "target_evidence": "green notebook", + "expected_visual_elements": [ + "sticky notes", + "green notebook" + ], + "negative_prompt": "blurred objects, overlapping elements, poor lighting, low detail", + "diversity_tags": [ + "similar-shape distractors", + "distinct colors" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "a3a3c294f30d", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic bulletin board with", + "max_similarity_to_other_prompt": 0.4, + "diversity_score": 0.4, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_a3a3c294f30d", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 21449348, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 3614937, + "qwen_elapsed_seconds": 5.171, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..c383521a2af315323b1791b0448c21f5b0b4dce7 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8b27312710048caa8f8473ebe56d04b6f02eae72506dd2f7b08cd45e9415e6d8 +size 476119 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..b312b351857aad63e1bc1927076916318c66b78a --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8abec7b08d35d6c7d7025d3c84e9e827f7e5b4e2721e4a0f42d65f89834b82e0 +size 624647 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..edbcc3d972bcea35b099a0a0efb7d5dfc544bd56 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:286fe35e76b3ca6d82f23c2166bb09cbd4344ac269ca82ac65c1db9776b62f61 +size 545427 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/record.json new file mode 100644 index 0000000000000000000000000000000000000000..15886698138aed9bdea20f71b13a060752771d4e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_a4819ff5cb89", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic locker filled with 40 small screws, one large silver screw stands out with sharp focus and clean edges", + "target_evidence": "large silver screw", + "expected_visual_elements": [ + "small screws", + "large silver screw" + ], + "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", + "diversity_tags": [ + "similar-scale distractors", + "sharp focus" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "a4819ff5cb89", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic locker filled with", + "max_similarity_to_other_prompt": 0.5417, + "diversity_score": 0.2583, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_a4819ff5cb89", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 27461534, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7471570, + "qwen_elapsed_seconds": 5.167, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..eba17732a3c95cbb31805dc2c4e53c38050d2ae4 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b45d7fad3f8c1f343d6c8aef367f8245861d6487eaaf058b764bae38279a2d0 +size 119509 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..5318556b8cca422163e5ffe866b0b6ef164d00b2 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b69c90f582946ad8d23d53fa4b48e78925b3b6c346df9d366518a0df4d8414a6 +size 691966 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..e972c6d89fdb9d0f5beadf6ff8520673a2a9c099 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0c3019eb1eaa537232eb360750d1d229ef95a2cf29267efab9ab63dca441c01 +size 563724 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/record.json new file mode 100644 index 0000000000000000000000000000000000000000..dbbc70bc8344bcf968da9976f0c1e393cf0de03f --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_a675e91f33b7", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic workbench with a single blue screw visible among 40 similar-shaped metallic items, high contrast and clear boundaries", + "target_evidence": "blue screw", + "expected_visual_elements": [ + "metallic items", + "blue screw" + ], + "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", + "diversity_tags": [ + "similar-shape distractors", + "high contrast" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "a675e91f33b7", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic workbench with a", + "max_similarity_to_other_prompt": 0.56, + "diversity_score": 0.24, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_a675e91f33b7", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_a675e91f33b7/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20985734, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7346072, + "qwen_elapsed_seconds": 5.16, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..8330d92d2df1984c8f693b497538451439ca72d6 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4d332662342e0cf7e04034e4fc4489a6cfcc4c2ef14494b50623ff84b1bafc9 +size 573699 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..55c5bbbd5fac11ec1f6edffe57aa34b69b8da2a3 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd8521cad25398a8b11ed21cee4dff5531a0c5d093c80f0f57f766ab4ab62625 +size 542238 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..9082cb8c53e7ad41f05c1b9a7b9e766863a8f982 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fcffe86a77ec927d26c725fd547d92c433faa72cf7e2193d5e22c02dd1ab758e +size 574010 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/record.json new file mode 100644 index 0000000000000000000000000000000000000000..c562829cb93e4cdf33a7ebb103f554d4c3c4e5cf --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_a722baad73c5", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "synthetic photorealistic market display with a red rubber duck among dozens of colorful balloons, clean boundaries, sharp focus on duck", + "target_evidence": "red rubber duck", + "expected_visual_elements": [ + "balloons", + "market display" + ], + "negative_prompt": "blurry focus, low contrast, overlapping objects", + "diversity_tags": [ + "color similarity", + "object scale variation" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "a722baad73c5", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.56, + "diversity_score": 0.24, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_a722baad73c5", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_a722baad73c5/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29163573, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6938740, + "qwen_elapsed_seconds": 5.184, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..e441c6d498b1981390b016b336952d2f5abbd70c --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cdb6c1949ba658f981bd84dcc44de062dd67466c961562d9fb4bbebe33536cec +size 484976 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..bc63c80026733234605d2bc86ba154baf01c8039 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf0b52ec5c56fdba5f73502bd785b8ba30a7e916936e0e65d90332e5a6ffbff5 +size 572369 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..3994916baf15ff6f409a3021f1a24a99207cc8fa --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb97d47e0d053b05bb1a0139082a95aae022897cbdabf0a3d02e1b4a4d938df1 +size 730116 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/record.json new file mode 100644 index 0000000000000000000000000000000000000000..c44e3c775062875286110fdc1048d77861a78dea --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_b0211aa536ef", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic market display with a single red key visible among 40 similar-shaped plastic items, clean boundaries, sharp focus", + "target_evidence": "single red key", + "expected_visual_elements": [ + "plastic items", + "red key" + ], + "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", + "diversity_tags": [ + "similar-shape distractors", + "high contrast" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "b0211aa536ef", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic market display with", + "max_similarity_to_other_prompt": 0.5385, + "diversity_score": 0.2615, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_b0211aa536ef", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_b0211aa536ef/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29478110, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1585783, + "qwen_elapsed_seconds": 5.114, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..5235d323871464e017916d1f4732e24dd2473f9e --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c84f7659f19c2822a095a28b57106211e03859af10852842727ce92b01ef199 +size 537259 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..44e500b02bc8bf3bef461d0bfec599bb1ef8e921 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:84592f6c6d0b57ace61ff4f7202586442904e6d36a6ec2e262eac0340c24ade4 +size 546841 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..e6f7fe4490a402001a766a1541a8634435d530fd --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:221323b25287dae234ac4485a2fec57df1d160d5877eb7979d494296fef9abb7 +size 559329 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/record.json new file mode 100644 index 0000000000000000000000000000000000000000..4465e7eb3c463cd2b4c02d893218411c782dfec0 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "clutter_distractor_d5_d1ee86278851", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic shelf with a single green book standing out among 40 similar-sized paperbacks, sharp focus and clean edges", + "target_evidence": "green book", + "expected_visual_elements": [ + "paperbacks", + "green book" + ], + "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", + "diversity_tags": [ + "similar-scale distractors", + "sharp focus" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "d1ee86278851", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic shelf with a", + "max_similarity_to_other_prompt": 0.3929, + "diversity_score": 0.4071, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_d1ee86278851", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_d1ee86278851/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 27530094, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6748699, + "qwen_elapsed_seconds": 5.126, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..c7fe28f15d34979ab6933e9b399280e7e1509254 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b7f5f7ba810804102f3ac847a12035144993c6c87db119483b51bf04d7088b9c +size 493676 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..00df91cd1ff118a572f99744a930065e403ffb6d --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d0ed7ec54a1ffe46347746bc61ed1d2dae31f0459f3216dd8e06c00ec5bb989 +size 600231 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..f7541b62c22be6e6c5cf1033c306ca14a02d6c54 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2ffe15ec7c2c60aeca08fae4bce9f11f64a021053bbc530aea7a74547a3165c +size 592167 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/record.json new file mode 100644 index 0000000000000000000000000000000000000000..af03faadced039050232fa167dc88e68bdfb3d07 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d5_dbb61cceb7f8", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "synthetic photorealistic shelf with a blue key surrounded by similar-shaped keys, clean foreground separation, sharp focus on target", + "target_evidence": "blue key", + "expected_visual_elements": [ + "blue key", + "similar-shaped keys", + "shelf" + ], + "negative_prompt": "overlapping keys, low contrast, poor focus, blurry background", + "diversity_tags": [ + "shape similarity", + "key clutter" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "dbb61cceb7f8", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic shelf with a", + "max_similarity_to_other_prompt": 0.6087, + "diversity_score": 0.1913, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_dbb61cceb7f8", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24433138, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 3994459, + "qwen_elapsed_seconds": 5.111, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/record.json" +} diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_00.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..830e2663fdaf6c58021a465b6bb1363ad16db153 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d596c9b0b29b12e38c897cc8126a4a0acdd0106e3e5699fe29143ad16c9beb3a +size 344388 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_01.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..7a4799ba3a9ce513f30672884120a43294537ed8 --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:565559b8fc98c39948627fb9b44b00b00d02ed79ef074c3d9a3eadd15a00a49a +size 613072 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_02.png b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..85ee69397dd8653ac93b691a687a0200d4ec682a --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b1f10e1f407776f10e632782dbb2a061ca52c12776e8921613a4c1039a0c4355 +size 546480 diff --git a/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/record.json b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/record.json new file mode 100644 index 0000000000000000000000000000000000000000..f0b7779132050fff9f6444b8aa720abc8144e3de --- /dev/null +++ b/vec_pilot/layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "clutter_distractor_d5_f953bafa1186", + "run_name": "flux2_qil_decomposition_pilot", + "category": "clutter_distractor", + "difficulty": 5, + "image_prompt": "synthetic photorealistic workbench with a green pen among many pens, sharp focus on target, clean boundaries, high contrast", + "target_evidence": "green pen", + "expected_visual_elements": [ + "green pen", + "many pens", + "workbench" + ], + "negative_prompt": "overlapping pens, low contrast, poor focus, blurry background", + "diversity_tags": [ + "color similarity", + "office items" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "f953bafa1186", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic workbench with a", + "max_similarity_to_other_prompt": 0.5652, + "diversity_score": 0.2348, + "repeated_opening": true, + "review": { + "prompt_id": "clutter_distractor_d5_f953bafa1186", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_f953bafa1186/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28371434, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_00.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_01.png", + "layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6821332, + "qwen_elapsed_seconds": 5.155, + "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/record.json" +} diff --git a/vec_pilot/layers-x/k3/manifest.jsonl b/vec_pilot/layers-x/k3/manifest.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..985cf33cc88c83d23ee6d4a6e97b97656293fbf4 --- /dev/null +++ b/vec_pilot/layers-x/k3/manifest.jsonl @@ -0,0 +1,591 @@ +{"prompt_id": "clutter_distractor_d1_19c3a4ffe857", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic market display with a yellow flashlight in the middle, surrounded by identical white flashlights and mirrors, clean edges for clarity", "target_evidence": "yellow flashlight", "expected_visual_elements": ["yellow flashlight", "white flashlights", "mirrors"], "negative_prompt": "low contrast, overlapping objects, watermarks", "diversity_tags": ["color similarity", "layout variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "19c3a4ffe857", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.3438, "diversity_score": 0.4562, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_19c3a4ffe857", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_19c3a4ffe857/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24711962, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_19c3a4ffe857/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_19c3a4ffe857/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_19c3a4ffe857/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9354240, "qwen_elapsed_seconds": 5.186, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_19c3a4ffe857/record.json"} +{"prompt_id": "clutter_distractor_d1_1ae6257850c8", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic tabletop with a green eraser on top, surrounded by other erasers and pencils, high contrast between eraser and background", "target_evidence": "green eraser", "expected_visual_elements": ["green eraser", "erasers", "pencils"], "negative_prompt": "low detail, watermarks, poor lighting", "diversity_tags": ["shape similarity", "object scale"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1ae6257850c8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tabletop with a", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_1ae6257850c8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_1ae6257850c8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22221330, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_1ae6257850c8/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_1ae6257850c8/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_1ae6257850c8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4709098, "qwen_elapsed_seconds": 5.164, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_1ae6257850c8/record.json"} +{"prompt_id": "clutter_distractor_d1_310a571d70b2", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic workbench with a silver stapler at the center, surrounded by other staplers and paper clips, clean background for clear separation", "target_evidence": "silver stapler", "expected_visual_elements": ["silver stapler", "staplers", "paper clips"], "negative_prompt": "low resolution, overlapping objects, watermarks", "diversity_tags": ["color similarity", "layout variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "310a571d70b2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.4667, "diversity_score": 0.3333, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_310a571d70b2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_310a571d70b2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26132914, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_310a571d70b2/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_310a571d70b2/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_310a571d70b2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6417602, "qwen_elapsed_seconds": 5.15, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_310a571d70b2/record.json"} +{"prompt_id": "clutter_distractor_d1_3a86f1190e0d", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic locker with a blue keychain standing out among keys, small screws, and lockpicks, high contrast edges", "target_evidence": "blue keychain", "expected_visual_elements": ["blue keychain", "keys", "screws", "lockpicks"], "negative_prompt": "low detail, overlapping items, dim lighting, unclear boundaries", "diversity_tags": ["color similarity", "object placement"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "3a86f1190e0d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_3a86f1190e0d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_3a86f1190e0d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26594641, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_3a86f1190e0d/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_3a86f1190e0d/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_3a86f1190e0d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9123108, "qwen_elapsed_seconds": 5.184, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_3a86f1190e0d/record.json"} +{"prompt_id": "clutter_distractor_d1_3f8c7a922e2b", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic workbench with a red wrench surrounded by identical blue bolts, screws, and nuts, clean boundaries for clear separation", "target_evidence": "red wrench", "expected_visual_elements": ["red wrench", "blue bolts", "screws", "nuts"], "negative_prompt": "blurry background, low contrast, overlapping objects, watermarks", "diversity_tags": ["color similarity", "object scale variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "3f8c7a922e2b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_3f8c7a922e2b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_3f8c7a922e2b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24134990, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_3f8c7a922e2b/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_3f8c7a922e2b/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_3f8c7a922e2b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2904303, "qwen_elapsed_seconds": 5.187, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_3f8c7a922e2b/record.json"} +{"prompt_id": "clutter_distractor_d1_4070f152f6d2", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic shelf with a white cup amidst bottles, jars, and containers, clear boundaries, sharp focus, high contrast lighting", "target_evidence": "white cup", "expected_visual_elements": ["white cup", "bottles", "jars", "containers"], "negative_prompt": "low resolution, blurry edges, overlapping objects, watermarks, logos", "diversity_tags": ["object scale variation", "layout diversity"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4070f152f6d2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.4444, "diversity_score": 0.3556, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_4070f152f6d2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_4070f152f6d2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27215966, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_4070f152f6d2/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_4070f152f6d2/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_4070f152f6d2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1652354, "qwen_elapsed_seconds": 5.165, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_4070f152f6d2/record.json"} +{"prompt_id": "clutter_distractor_d1_48c922f0d5ae", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic market display with a red pen amidst similar-colored items, sharp focus on the red one", "target_evidence": "red pen", "expected_visual_elements": ["red pen", "blue pens", "black pens", "pencils"], "negative_prompt": "blurry elements, low contrast, occluded items, background clutter", "diversity_tags": ["color similarity", "object placement"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "48c922f0d5ae", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_48c922f0d5ae", "accept": true, "quality": 4, "decomposition_fit": 4, "diversity": 3, "issues": ["Red pen among similar-colored items may reduce visual distinctiveness; potential for low contrast"]}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.84, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_48c922f0d5ae/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29182245, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_48c922f0d5ae/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_48c922f0d5ae/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_48c922f0d5ae/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9205599, "qwen_elapsed_seconds": 5.166, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_48c922f0d5ae/record.json"} +{"prompt_id": "clutter_distractor_d1_5f1fb5e2b9b6", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic tabletop with a single orange pen amidst many black pens, sharp focus on the orange one", "target_evidence": "orange pen", "expected_visual_elements": ["orange pen", "black pens", "notebooks", "folders"], "negative_prompt": "low detail, overlapping objects, poor lighting, background clutter", "diversity_tags": ["color contrast", "object placement"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5f1fb5e2b9b6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tabletop with a", "max_similarity_to_other_prompt": 0.5217, "diversity_score": 0.2783, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_5f1fb5e2b9b6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_5f1fb5e2b9b6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26965689, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_5f1fb5e2b9b6/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_5f1fb5e2b9b6/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_5f1fb5e2b9b6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9129296, "qwen_elapsed_seconds": 5.107, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_5f1fb5e2b9b6/record.json"} +{"prompt_id": "clutter_distractor_d1_71064c32726c", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic workbench with a blue pencil on the left, surrounded by other pencils and erasers, high contrast for clear separation", "target_evidence": "blue pencil", "expected_visual_elements": ["blue pencil", "pencils", "erasers"], "negative_prompt": "low detail, overlapping objects, watermarks", "diversity_tags": ["color similarity", "layout variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "71064c32726c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_71064c32726c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_71064c32726c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21836076, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_71064c32726c/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_71064c32726c/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_71064c32726c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1887605, "qwen_elapsed_seconds": 5.19, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_71064c32726c/record.json"} +{"prompt_id": "clutter_distractor_d1_8f2b38de4366", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic locker with a green book among tangled cables, loose screws, and small plastic parts, clean boundaries, sharp focus", "target_evidence": "green book", "expected_visual_elements": ["green book", "tangled cables", "loose screws", "small plastic parts"], "negative_prompt": "low resolution, blurry edges, overlapping objects, watermarks, logos", "diversity_tags": ["color similarity", "object density variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "8f2b38de4366", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.4074, "diversity_score": 0.3926, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_8f2b38de4366", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_8f2b38de4366/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29984344, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_8f2b38de4366/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_8f2b38de4366/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_8f2b38de4366/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 486178, "qwen_elapsed_seconds": 5.176, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_8f2b38de4366/record.json"} +{"prompt_id": "clutter_distractor_d1_9bb2e5a22e61", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic locker with a black eraser at the center, surrounded by other erasers and rulers, clear foreground separation for easy identification", "target_evidence": "black eraser", "expected_visual_elements": ["black eraser", "erasers", "rulers"], "negative_prompt": "low resolution, poor lighting, watermarks", "diversity_tags": ["shape similarity", "object scale"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "9bb2e5a22e61", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.4667, "diversity_score": 0.3333, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_9bb2e5a22e61", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_9bb2e5a22e61/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24959664, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_9bb2e5a22e61/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_9bb2e5a22e61/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_9bb2e5a22e61/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 633246, "qwen_elapsed_seconds": 5.171, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_9bb2e5a22e61/record.json"} +{"prompt_id": "clutter_distractor_d1_a8ee7e503542", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic bulletin board with a green pen at the center, surrounded by numerous yellow sticky notes and paper clips, sharp focus on the pen", "target_evidence": "green pen", "expected_visual_elements": ["green pen", "yellow sticky notes", "paper clips"], "negative_prompt": "low resolution, overlapping text, out-of-focus elements", "diversity_tags": ["shape similarity", "layout variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a8ee7e503542", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.4667, "diversity_score": 0.3333, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_a8ee7e503542", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_a8ee7e503542/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24844628, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_a8ee7e503542/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_a8ee7e503542/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_a8ee7e503542/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 367013, "qwen_elapsed_seconds": 5.158, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_a8ee7e503542/record.json"} +{"prompt_id": "clutter_distractor_d1_c39c47d316f9", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic market display with a red notebook among scattered pens, paper clips, and rulers, sharp focus on the notebook's clean edges", "target_evidence": "red notebook", "expected_visual_elements": ["red notebook", "scattered pens", "paper clips", "rulers"], "negative_prompt": "blurry background, low contrast, overlapping objects, unrealistic lighting", "diversity_tags": ["color contrast", "object scale variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c39c47d316f9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.4483, "diversity_score": 0.3517, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_c39c47d316f9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_c39c47d316f9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29463184, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_c39c47d316f9/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_c39c47d316f9/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_c39c47d316f9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4858022, "qwen_elapsed_seconds": 5.243, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_c39c47d316f9/record.json"} +{"prompt_id": "clutter_distractor_d1_cc52f08cb218", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic workbench with a blue pen surrounded by screws, wires, and small gears, clear foreground separation, high contrast lighting", "target_evidence": "blue pen", "expected_visual_elements": ["blue pen", "screws", "wires", "small gears"], "negative_prompt": "low resolution, blurry edges, overlapping objects, watermarks, logos", "diversity_tags": ["shape similarity", "layout variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cc52f08cb218", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_cc52f08cb218", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_cc52f08cb218/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22574722, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_cc52f08cb218/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_cc52f08cb218/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_cc52f08cb218/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2240079, "qwen_elapsed_seconds": 5.115, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_cc52f08cb218/record.json"} +{"prompt_id": "clutter_distractor_d1_cc5a43d620ae", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic shelf with a red pen on the middle, surrounded by similar-colored pens and markers, sharp focus on the pen", "target_evidence": "red pen", "expected_visual_elements": ["red pen", "similar-colored pens", "markers"], "negative_prompt": "low contrast, overlapping objects, watermarks", "diversity_tags": ["color similarity", "layout variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cc5a43d620ae", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.64, "diversity_score": 0.16, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_cc5a43d620ae", "accept": true, "quality": 4, "decomposition_fit": 4, "diversity": 3, "issues": ["Red pen among similar-colored pens and markers may reduce visual distinctiveness; potential for low contrast"]}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.84, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_cc5a43d620ae/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20370453, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_cc5a43d620ae/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_cc5a43d620ae/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_cc5a43d620ae/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3203227, "qwen_elapsed_seconds": 5.172, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_cc5a43d620ae/record.json"} +{"prompt_id": "clutter_distractor_d1_ce8be3e19d8a", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic market display with a red key on top of scattered coins, bottle caps, and buttons, all in high contrast, clean boundaries, sharp focus", "target_evidence": "red key", "expected_visual_elements": ["red key", "scattered coins", "bottle caps", "buttons"], "negative_prompt": "low resolution, blurry edges, overlapping objects, watermarks, logos", "diversity_tags": ["color similarity", "object scale variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ce8be3e19d8a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.5152, "diversity_score": 0.2848, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_ce8be3e19d8a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_ce8be3e19d8a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26160142, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_ce8be3e19d8a/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_ce8be3e19d8a/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_ce8be3e19d8a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6070860, "qwen_elapsed_seconds": 5.204, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_ce8be3e19d8a/record.json"} +{"prompt_id": "clutter_distractor_d1_eaf963085883", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic shelf with a white key on top, surrounded by various keys and lockers, high contrast between key and background", "target_evidence": "white key", "expected_visual_elements": ["white key", "various keys", "lockers"], "negative_prompt": "low detail, watermarks, poor lighting", "diversity_tags": ["object scale", "shape similarity"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "eaf963085883", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_eaf963085883", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_eaf963085883/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21530217, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_eaf963085883/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_eaf963085883/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_eaf963085883/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8782574, "qwen_elapsed_seconds": 5.159, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_eaf963085883/record.json"} +{"prompt_id": "clutter_distractor_d1_f0a73775374e", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic bulletin board with a yellow note on top of sticky notes, paper scraps, and clips, high contrast, clean foreground separation", "target_evidence": "yellow note", "expected_visual_elements": ["yellow note", "sticky notes", "paper scraps", "clips"], "negative_prompt": "low resolution, blurry edges, overlapping objects, watermarks, logos", "diversity_tags": ["shape similarity", "text presence"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f0a73775374e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.6538, "diversity_score": 0.1462, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_f0a73775374e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_f0a73775374e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24154012, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_f0a73775374e/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_f0a73775374e/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_f0a73775374e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5894778, "qwen_elapsed_seconds": 5.199, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_f0a73775374e/record.json"} +{"prompt_id": "clutter_distractor_d1_f4acb3edd9d2", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 1, "image_prompt": "synthetic photorealistic locker with a blue ruler on the top, surrounded by rulers and measuring tapes, sharp focus on the ruler", "target_evidence": "blue ruler", "expected_visual_elements": ["blue ruler", "rulers", "measuring tapes"], "negative_prompt": "blurry elements, poor lighting, watermarks", "diversity_tags": ["shape similarity", "object scale"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f4acb3edd9d2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.44, "diversity_score": 0.36, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d1_f4acb3edd9d2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d1_f4acb3edd9d2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21717013, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d1_f4acb3edd9d2/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_f4acb3edd9d2/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d1_f4acb3edd9d2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7758025, "qwen_elapsed_seconds": 5.191, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d1_f4acb3edd9d2/record.json"} +{"prompt_id": "clutter_distractor_d2_033799c2b1fc", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "Synthetic photorealistic market display with a red wrench surrounded by 10 similar-colored tools, clean boundaries, and sharp focus on the wrench.", "target_evidence": "red wrench", "expected_visual_elements": ["red wrench", "similar-colored tools"], "negative_prompt": "blurry background, overlapping objects, low contrast", "diversity_tags": ["similar-color distractors", "market display"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "033799c2b1fc", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.7727, "diversity_score": 0.0273, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_033799c2b1fc", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_033799c2b1fc/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27386340, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_033799c2b1fc/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_033799c2b1fc/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_033799c2b1fc/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6142064, "qwen_elapsed_seconds": 5.228, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_033799c2b1fc/record.json"} +{"prompt_id": "clutter_distractor_d2_0a183eb4b3c5", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic bulletin board with a yellow sticky note above multiple blue notes, sharp focus on yellow note", "target_evidence": "yellow sticky note", "expected_visual_elements": ["bulletin board", "blue notes", "paper scraps", "pins"], "negative_prompt": "blurry, low contrast, overlapping elements", "diversity_tags": ["color similarity", "text region"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0a183eb4b3c5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.5909, "diversity_score": 0.2091, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_0a183eb4b3c5", "accept": true, "quality": 4, "decomposition_fit": 4, "diversity": 3, "issues": ["repetitive scene: bulletin board with sticky notes"]}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.84, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_0a183eb4b3c5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24553536, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_0a183eb4b3c5/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_0a183eb4b3c5/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_0a183eb4b3c5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5380664, "qwen_elapsed_seconds": 5.13, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_0a183eb4b3c5/record.json"} +{"prompt_id": "clutter_distractor_d2_136a3d9f7946", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic locker with a white phone case among black phones, chargers, and cables, sharp focus on phone case", "target_evidence": "white phone case", "expected_visual_elements": ["locker", "black phones", "chargers", "cables"], "negative_prompt": "blurry, low contrast, overlapping objects", "diversity_tags": ["color similarity", "object scale"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "136a3d9f7946", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.3704, "diversity_score": 0.4296, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_136a3d9f7946", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_136a3d9f7946/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28070631, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_136a3d9f7946/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_136a3d9f7946/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_136a3d9f7946/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9998223, "qwen_elapsed_seconds": 5.152, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_136a3d9f7946/record.json"} +{"prompt_id": "clutter_distractor_d2_23fe340492f0", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic market display with a blue marker amidst various pens, highlighters, and colored pencils, all in clean boundaries and clear foreground", "target_evidence": "blue marker", "expected_visual_elements": ["blue marker", "pens", "highlighters", "colored pencils"], "negative_prompt": "blurry, low resolution, poor lighting, overexposed", "diversity_tags": ["color similarity", "object scale variation", "clean boundaries"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "23fe340492f0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.4138, "diversity_score": 0.3862, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_23fe340492f0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_23fe340492f0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25666010, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_23fe340492f0/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_23fe340492f0/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_23fe340492f0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3407511, "qwen_elapsed_seconds": 5.153, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_23fe340492f0/record.json"} +{"prompt_id": "clutter_distractor_d2_2864d2d7a640", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic workbench with a single red wrench amidst scattered screws, bolts, and tools, clean boundaries and high contrast for clear separation", "target_evidence": "single red wrench", "expected_visual_elements": ["red wrench", "scattered screws and bolts", "tools"], "negative_prompt": "blurry elements, overlapping objects, low contrast, watermarks, text, logos", "diversity_tags": ["tool clutter", "color contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2864d2d7a640", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_2864d2d7a640", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_2864d2d7a640/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29120783, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_2864d2d7a640/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_2864d2d7a640/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_2864d2d7a640/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8226286, "qwen_elapsed_seconds": 5.204, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_2864d2d7a640/record.json"} +{"prompt_id": "clutter_distractor_d2_2eb6156bd2b2", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic tabletop with a white ruler among various rulers, measuring tapes, and protractors, all in high contrast and clear foreground", "target_evidence": "white ruler", "expected_visual_elements": ["white ruler", "rulers", "measuring tapes", "protractors"], "negative_prompt": "blurry, low resolution, poor lighting, overexposed", "diversity_tags": ["shape similarity", "object scale variation", "clear foreground"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2eb6156bd2b2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tabletop with a", "max_similarity_to_other_prompt": 0.4074, "diversity_score": 0.3926, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_2eb6156bd2b2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_2eb6156bd2b2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28612875, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_2eb6156bd2b2/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_2eb6156bd2b2/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_2eb6156bd2b2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 658631, "qwen_elapsed_seconds": 5.209, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_2eb6156bd2b2/record.json"} +{"prompt_id": "clutter_distractor_d2_43b3b86afc18", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic locker with a black pen standing out against white erasers, sticky notes, and small markers, all in sharp focus and high contrast", "target_evidence": "black pen", "expected_visual_elements": ["black pen", "white erasers", "sticky notes", "markers"], "negative_prompt": "blurry, low resolution, poor lighting, overexposed", "diversity_tags": ["color similarity", "object scale variation", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "43b3b86afc18", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.4062, "diversity_score": 0.3937, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_43b3b86afc18", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_43b3b86afc18/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24034344, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_43b3b86afc18/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_43b3b86afc18/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_43b3b86afc18/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9416524, "qwen_elapsed_seconds": 5.096, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_43b3b86afc18/record.json"} +{"prompt_id": "clutter_distractor_d2_4ed1ed4d6471", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic bulletin board with a purple lightbulb placed among different colored balloons, strings, and paper cutouts, all in sharp focus and clean separation", "target_evidence": "purple lightbulb", "expected_visual_elements": ["purple lightbulb", "balloons", "strings", "paper cutouts"], "negative_prompt": "blurry, low resolution, poor lighting, overexposed", "diversity_tags": ["color similarity", "object scale variation", "clean separation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4ed1ed4d6471", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.4062, "diversity_score": 0.3937, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_4ed1ed4d6471", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_4ed1ed4d6471/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29399899, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_4ed1ed4d6471/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_4ed1ed4d6471/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_4ed1ed4d6471/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7772610, "qwen_elapsed_seconds": 5.147, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_4ed1ed4d6471/record.json"} +{"prompt_id": "clutter_distractor_d2_5667b5cb027c", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic workbench with a green pen placed among different colored pens, rulers, and small clips, all in high contrast and sharp focus", "target_evidence": "green pen", "expected_visual_elements": ["green pen", "pens", "rulers", "clips"], "negative_prompt": "blurry, low resolution, poor lighting, overexposed", "diversity_tags": ["shape similarity", "object scale variation", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5667b5cb027c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_5667b5cb027c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_5667b5cb027c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23520963, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_5667b5cb027c/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_5667b5cb027c/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_5667b5cb027c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8990774, "qwen_elapsed_seconds": 5.148, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_5667b5cb027c/record.json"} +{"prompt_id": "clutter_distractor_d2_5c6b04bda464", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic locker with a yellow magnifying glass amidst scattered keys, buttons, and small gears, all in high contrast and clean boundaries", "target_evidence": "yellow magnifying glass", "expected_visual_elements": ["yellow magnifying glass", "keys", "buttons", "gears"], "negative_prompt": "blurry, low resolution, poor lighting, overexposed", "diversity_tags": ["shape similarity", "object scale variation", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5c6b04bda464", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_5c6b04bda464", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_5c6b04bda464/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21281373, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_5c6b04bda464/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_5c6b04bda464/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_5c6b04bda464/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3428654, "qwen_elapsed_seconds": 5.13, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_5c6b04bda464/record.json"} +{"prompt_id": "clutter_distractor_d2_5f28522c0497", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic locker with a green key hidden among 8 metallic items, high contrast lighting, clear foreground separation", "target_evidence": "green key", "expected_visual_elements": ["green key", "metallic items"], "negative_prompt": "low resolution, cluttered, overlapping elements, soft shadows", "diversity_tags": ["key", "shape similarity"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5f28522c0497", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.6818, "diversity_score": 0.1182, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_5f28522c0497", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_5f28522c0497/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28219616, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_5f28522c0497/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_5f28522c0497/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_5f28522c0497/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3914601, "qwen_elapsed_seconds": 5.099, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_5f28522c0497/record.json"} +{"prompt_id": "clutter_distractor_d2_62c41fa339e9", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic shelf with a red pen surrounded by similar-colored pens, markers, and small labels, all in clear foreground and sharp focus", "target_evidence": "red pen", "expected_visual_elements": ["red pen", "pens", "markers", "labels"], "negative_prompt": "blurry, low resolution, poor lighting, overexposed", "diversity_tags": ["color similarity", "object scale variation", "clear foreground"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "62c41fa339e9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.64, "diversity_score": 0.16, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_62c41fa339e9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_62c41fa339e9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25391625, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_62c41fa339e9/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_62c41fa339e9/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_62c41fa339e9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 400358, "qwen_elapsed_seconds": 5.148, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_62c41fa339e9/record.json"} +{"prompt_id": "clutter_distractor_d2_71e196099798", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic workbench with a red wrench surrounded by 10 similar tools, clean boundaries, sharp focus on the wrench", "target_evidence": "red wrench", "expected_visual_elements": ["red wrench", "tools"], "negative_prompt": "blurry, low contrast, out-of-focus, over-saturated", "diversity_tags": ["tool", "color contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "71e196099798", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.7727, "diversity_score": 0.0273, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_71e196099798", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_71e196099798/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25914364, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_71e196099798/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_71e196099798/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_71e196099798/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2681785, "qwen_elapsed_seconds": 5.154, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_71e196099798/record.json"} +{"prompt_id": "clutter_distractor_d2_8e38d5da6fb1", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "Synthetic photorealistic locker with a green key nestled among 12 metallic objects, high contrast lighting, and clear foreground separation.", "target_evidence": "green key", "expected_visual_elements": ["green key", "metallic objects"], "negative_prompt": "low resolution, out-of-focus elements, poor lighting", "diversity_tags": ["similar-shape distractors", "locker scene"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "8e38d5da6fb1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.6818, "diversity_score": 0.1182, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_8e38d5da6fb1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_8e38d5da6fb1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27429348, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_8e38d5da6fb1/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_8e38d5da6fb1/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_8e38d5da6fb1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8249001, "qwen_elapsed_seconds": 5.169, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_8e38d5da6fb1/record.json"} +{"prompt_id": "clutter_distractor_d2_8eae6d7ab95d", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic bulletin board with a yellow sticky note above multiple blue notes, paper scraps, and pins, clean boundaries", "target_evidence": "yellow sticky note", "expected_visual_elements": ["bulletin board", "blue notes", "paper scraps", "pins"], "negative_prompt": "blurry, overlapping, low contrast", "diversity_tags": ["color similarity", "text region"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "8eae6d7ab95d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.5909, "diversity_score": 0.2091, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_8eae6d7ab95d", "accept": true, "quality": 4, "decomposition_fit": 4, "diversity": 3, "issues": ["repetitive scene: bulletin board with sticky notes and added clutter"]}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.84, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_8eae6d7ab95d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20815220, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_8eae6d7ab95d/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_8eae6d7ab95d/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_8eae6d7ab95d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2137788, "qwen_elapsed_seconds": 5.195, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_8eae6d7ab95d/record.json"} +{"prompt_id": "clutter_distractor_d2_93ba85ec2f43", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic tabletop with a green mug beside coffee cups, mugs, and cups, high contrast and clear boundaries", "target_evidence": "green mug", "expected_visual_elements": ["tabletop", "coffee cups", "mugs", "cups"], "negative_prompt": "low resolution, overlapping, out of focus", "diversity_tags": ["color similarity", "object scale"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "93ba85ec2f43", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tabletop with a", "max_similarity_to_other_prompt": 0.4167, "diversity_score": 0.3833, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_93ba85ec2f43", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_93ba85ec2f43/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27084897, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_93ba85ec2f43/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_93ba85ec2f43/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_93ba85ec2f43/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9867592, "qwen_elapsed_seconds": 5.171, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_93ba85ec2f43/record.json"} +{"prompt_id": "clutter_distractor_d2_d090f79ec7e8", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic workbench with a blue USB drive placed among scattered paper clips, rubber bands, and small screws, all in clean boundaries and sharp focus", "target_evidence": "blue USB drive", "expected_visual_elements": ["blue USB drive", "paper clips", "rubber bands", "screws"], "negative_prompt": "blurry, low resolution, poor lighting, overexposed", "diversity_tags": ["shape similarity", "object scale variation", "clean boundaries"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d090f79ec7e8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.4375, "diversity_score": 0.3625, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_d090f79ec7e8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_d090f79ec7e8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21108020, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_d090f79ec7e8/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_d090f79ec7e8/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_d090f79ec7e8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 118401, "qwen_elapsed_seconds": 5.188, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_d090f79ec7e8/record.json"} +{"prompt_id": "clutter_distractor_d2_d8cebd503166", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic locker with a red keychain amidst blue pens, sticky notes, and crumpled paper, sharp focus on keychain", "target_evidence": "red keychain", "expected_visual_elements": ["locker", "blue pens", "sticky notes", "crumpled paper"], "negative_prompt": "blurry, low contrast, overlapping objects, low resolution", "diversity_tags": ["color similarity", "object scale variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d8cebd503166", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.4167, "diversity_score": 0.3833, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_d8cebd503166", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_d8cebd503166/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26716341, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_d8cebd503166/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_d8cebd503166/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_d8cebd503166/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7140709, "qwen_elapsed_seconds": 5.088, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_d8cebd503166/record.json"} +{"prompt_id": "clutter_distractor_d2_d9757fba4684", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic market display with a red keychain surrounded by colorful bottle caps, loose coins, and small plastic containers, all in high contrast and sharp focus", "target_evidence": "red keychain", "expected_visual_elements": ["red keychain", "colorful bottle caps", "loose coins", "plastic containers"], "negative_prompt": "blurry, low resolution, poor lighting, overexposed", "diversity_tags": ["color similarity", "object scale variation", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d9757fba4684", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.5152, "diversity_score": 0.2848, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_d9757fba4684", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_d9757fba4684/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28011191, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_d9757fba4684/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_d9757fba4684/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_d9757fba4684/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8396121, "qwen_elapsed_seconds": 5.094, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_d9757fba4684/record.json"} +{"prompt_id": "clutter_distractor_d2_f0e23f1a589a", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 2, "image_prompt": "synthetic photorealistic shelf with a green notebook standing out among various pens, rulers, and sticky notes, all in sharp focus and clear foreground separation", "target_evidence": "green notebook", "expected_visual_elements": ["green notebook", "pens", "rulers", "sticky notes"], "negative_prompt": "blurry, low resolution, poor lighting, overexposed", "diversity_tags": ["color similarity", "object scale variation", "clear foreground"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f0e23f1a589a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.4062, "diversity_score": 0.3937, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d2_f0e23f1a589a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d2_f0e23f1a589a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23116231, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d2_f0e23f1a589a/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_f0e23f1a589a/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d2_f0e23f1a589a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6843202, "qwen_elapsed_seconds": 5.172, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d2_f0e23f1a589a/record.json"} +{"prompt_id": "clutter_distractor_d3_040a39a09ccc", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic workbench with a small key hidden among screws, nuts, and tools, sharp focus on the key for clear identification", "target_evidence": "small key", "expected_visual_elements": ["key", "screws", "nuts", "tools"], "negative_prompt": "overlapping items, low resolution, out-of-focus", "diversity_tags": ["shape similarity", "object scale"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "040a39a09ccc", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_040a39a09ccc", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_040a39a09ccc/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28367993, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_040a39a09ccc/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_040a39a09ccc/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_040a39a09ccc/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6497521, "qwen_elapsed_seconds": 5.149, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_040a39a09ccc/record.json"} +{"prompt_id": "clutter_distractor_d3_0d8e22361944", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic locker with a red keychain among scattered keys, screws, and small tools, high contrast and clean boundaries", "target_evidence": "red keychain", "expected_visual_elements": ["locker", "keys", "screws", "tools"], "negative_prompt": "blurry focus, overlapping objects, low contrast, watermarks", "diversity_tags": ["color similarity", "object scale"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0d8e22361944", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_0d8e22361944", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_0d8e22361944/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25460222, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_0d8e22361944/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_0d8e22361944/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_0d8e22361944/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5656867, "qwen_elapsed_seconds": 5.151, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_0d8e22361944/record.json"} +{"prompt_id": "clutter_distractor_d3_126fb86645c8", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic market display with a red apple among green leaves, clean boundaries, sharp focus, and high contrast", "target_evidence": "red apple", "expected_visual_elements": ["market display", "green leaves", "red apple"], "negative_prompt": "blurry, low contrast, out-of-focus, poor lighting", "diversity_tags": ["color contrast", "natural objects"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "126fb86645c8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_126fb86645c8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_126fb86645c8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21178795, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_126fb86645c8/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_126fb86645c8/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_126fb86645c8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2716002, "qwen_elapsed_seconds": 5.14, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_126fb86645c8/record.json"} +{"prompt_id": "clutter_distractor_d3_17f4b0cf2f62", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic bulletin board with a green marker on top of similar markers and colored pencils, high contrast and clear separation", "target_evidence": "green marker", "expected_visual_elements": ["bulletin board", "markers", "colored pencils"], "negative_prompt": "low resolution, overlapping elements, watermarks", "diversity_tags": ["color similarity", "object placement"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "17f4b0cf2f62", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_17f4b0cf2f62", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24786734, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 565084, "qwen_elapsed_seconds": 5.099, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_17f4b0cf2f62/record.json"} +{"prompt_id": "clutter_distractor_d3_1fb8fe8d0e70", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic shelf with a blue bottle standing out among glasses, cups, and jars, clear boundaries for easy detection", "target_evidence": "blue bottle", "expected_visual_elements": ["bottle", "glasses", "cups", "jars"], "negative_prompt": "overlapping items, low contrast, out-of-focus", "diversity_tags": ["color similarity", "object scale"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1fb8fe8d0e70", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.3214, "diversity_score": 0.4786, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_1fb8fe8d0e70", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25913680, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3514110, "qwen_elapsed_seconds": 5.164, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_1fb8fe8d0e70/record.json"} +{"prompt_id": "clutter_distractor_d3_24fe198c21d6", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic workbench with a blue screwdriver surrounded by nuts, bolts, and wires, sharp focus and clear foreground separation", "target_evidence": "blue screwdriver", "expected_visual_elements": ["workbench", "nuts", "bolts", "wires"], "negative_prompt": "low resolution, cluttered background, watermarks", "diversity_tags": ["shape similarity", "layout variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "24fe198c21d6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_24fe198c21d6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_24fe198c21d6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25933561, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2257078, "qwen_elapsed_seconds": 5.197, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_24fe198c21d6/record.json"} +{"prompt_id": "clutter_distractor_d3_2a92bf5e61f6", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic bulletin board with a yellow sticky note among other notes, paper scraps, and markers, clean foreground separation", "target_evidence": "yellow sticky note", "expected_visual_elements": ["sticky note", "paper scraps", "markers", "post-it notes"], "negative_prompt": "blurred edges, mixed colors, overlapping notes", "diversity_tags": ["shape similarity", "text region"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2a92bf5e61f6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.64, "diversity_score": 0.16, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_2a92bf5e61f6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20905239, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8752243, "qwen_elapsed_seconds": 5.124, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_2a92bf5e61f6/record.json"} +{"prompt_id": "clutter_distractor_d3_4673f2297c2b", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic shelf with a white mug in a row of similar mugs, clean boundaries and high contrast for easy separation", "target_evidence": "white mug", "expected_visual_elements": ["shelf", "mugs", "similar objects"], "negative_prompt": "low resolution, overlapping elements, watermarks", "diversity_tags": ["shape similarity", "layout variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4673f2297c2b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_4673f2297c2b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_4673f2297c2b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29062757, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2220584, "qwen_elapsed_seconds": 5.166, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_4673f2297c2b/record.json"} +{"prompt_id": "clutter_distractor_d3_4f0d15b53f93", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic market display with a yellow calendar on a stack of similar colored items, sharp focus and clear separation", "target_evidence": "yellow calendar", "expected_visual_elements": ["market display", "stack of items", "similar colored objects"], "negative_prompt": "blurred text, overlapping objects, watermarks", "diversity_tags": ["color similarity", "object scale"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4f0d15b53f93", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_4f0d15b53f93", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22684801, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4249307, "qwen_elapsed_seconds": 5.171, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_4f0d15b53f93/record.json"} +{"prompt_id": "clutter_distractor_d3_56877d3370c6", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic locker with a purple eraser among many erasers and stationery, high contrast and clear foreground focus", "target_evidence": "purple eraser", "expected_visual_elements": ["locker", "erasers", "stationery"], "negative_prompt": "low detail, overlapping objects, watermarks", "diversity_tags": ["color similarity", "object placement"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "56877d3370c6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.44, "diversity_score": 0.36, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_56877d3370c6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_56877d3370c6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28241459, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2188272, "qwen_elapsed_seconds": 5.17, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_56877d3370c6/record.json"} +{"prompt_id": "clutter_distractor_d3_596c2e1f65c8", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic workbench with a red ruler among rulers, measuring tapes, and other tools, clean boundaries and sharp focus", "target_evidence": "red ruler", "expected_visual_elements": ["workbench", "rulers", "measuring tapes", "tools"], "negative_prompt": "blurry focus, overlapping elements, watermarks", "diversity_tags": ["shape similarity", "object scale"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "596c2e1f65c8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.44, "diversity_score": 0.36, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_596c2e1f65c8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29983416, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1344084, "qwen_elapsed_seconds": 5.119, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_596c2e1f65c8/record.json"} +{"prompt_id": "clutter_distractor_d3_5c9d227b32b4", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "Synthetic photorealistic bulletin board with a blue notebook amidst similar-shaped paper items, sharp focus on notebook, high contrast", "target_evidence": "blue notebook", "expected_visual_elements": ["paper items", "bulletin board"], "negative_prompt": "blurry items, low contrast, overlapping objects, poor lighting", "diversity_tags": ["similar-shape distractors", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5c9d227b32b4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.44, "diversity_score": 0.36, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_5c9d227b32b4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24649423, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1950292, "qwen_elapsed_seconds": 5.182, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_5c9d227b32b4/record.json"} +{"prompt_id": "clutter_distractor_d3_8c9c64b73906", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic market display with a red traffic cone amidst colorful balloons, confetti, and paper streamers, clean boundaries for easy separation", "target_evidence": "red traffic cone", "expected_visual_elements": ["traffic cone", "balloons", "confetti", "paper streamers"], "negative_prompt": "blurry objects, overlapping elements, low contrast", "diversity_tags": ["color similarity", "layout variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "8c9c64b73906", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.3793, "diversity_score": 0.4207, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_8c9c64b73906", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_8c9c64b73906/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28250007, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6699584, "qwen_elapsed_seconds": 5.142, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_8c9c64b73906/record.json"} +{"prompt_id": "clutter_distractor_d3_b05af666789f", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic market display with a blue pen in a group of similar pens and paper, sharp focus and clear foreground separation", "target_evidence": "blue pen", "expected_visual_elements": ["market display", "pens", "paper"], "negative_prompt": "blurred text, overlapping objects, watermarks", "diversity_tags": ["color similarity", "layout variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b05af666789f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_b05af666789f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_b05af666789f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21139338, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1276532, "qwen_elapsed_seconds": 5.164, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_b05af666789f/record.json"} +{"prompt_id": "clutter_distractor_d3_b0fda5852ae2", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "Synthetic photorealistic market display with a red rubber duck surrounded by similar-colored plastic toys, sharp focus on duck, high contrast edges", "target_evidence": "red rubber duck", "expected_visual_elements": ["plastic toys", "market display"], "negative_prompt": "blurry background, low contrast, overlapping objects, poor lighting", "diversity_tags": ["similar-color distractors", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b0fda5852ae2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_b0fda5852ae2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29488989, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9192459, "qwen_elapsed_seconds": 5.135, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_b0fda5852ae2/record.json"} +{"prompt_id": "clutter_distractor_d3_c3ccd586d6ae", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic shelf with a yellow book in a row of similar books, clean boundaries and high contrast for easy identification", "target_evidence": "yellow book", "expected_visual_elements": ["shelf", "books", "similar objects"], "negative_prompt": "low resolution, overlapping elements, watermarks", "diversity_tags": ["shape similarity", "object scale"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c3ccd586d6ae", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_c3ccd586d6ae", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27839720, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5146656, "qwen_elapsed_seconds": 5.101, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_c3ccd586d6ae/record.json"} +{"prompt_id": "clutter_distractor_d3_c685f2fb5896", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic bulletin board with a green pen on top of sticky notes, pens, and paper clips, high contrast and clean edges", "target_evidence": "green pen", "expected_visual_elements": ["bulletin board", "sticky notes", "pens", "paper clips"], "negative_prompt": "out of focus, overlapping elements, low detail", "diversity_tags": ["color similarity", "object placement"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c685f2fb5896", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.6538, "diversity_score": 0.1462, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_c685f2fb5896", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_c685f2fb5896/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26948933, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2772937, "qwen_elapsed_seconds": 5.227, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_c685f2fb5896/record.json"} +{"prompt_id": "clutter_distractor_d3_cbfa252c1b7d", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic tabletop with a white eraser among pencils, rulers, and paper clips, sharp focus on the eraser for clear identification", "target_evidence": "white eraser", "expected_visual_elements": ["eraser", "pencils", "rulers", "paper clips"], "negative_prompt": "low resolution, poor lighting, background clutter", "diversity_tags": ["shape similarity", "object scale"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cbfa252c1b7d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tabletop with a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_cbfa252c1b7d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23354054, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7591050, "qwen_elapsed_seconds": 5.203, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_cbfa252c1b7d/record.json"} +{"prompt_id": "clutter_distractor_d3_d2419f42b8a8", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "Synthetic photorealistic market display with a green bottle among similar-shaped glass containers, sharp focus on bottle, high contrast", "target_evidence": "green bottle", "expected_visual_elements": ["glass containers", "market display"], "negative_prompt": "blurry items, low contrast, overlapping objects, poor lighting", "diversity_tags": ["similar-shape distractors", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d2419f42b8a8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_d2419f42b8a8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21513032, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1374224, "qwen_elapsed_seconds": 5.105, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_d2419f42b8a8/record.json"} +{"prompt_id": "clutter_distractor_d3_eb19a4572a58", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 3, "image_prompt": "synthetic photorealistic locker with a green ID badge placed between keys, coins, and pens, high contrast for visual separation", "target_evidence": "green ID badge", "expected_visual_elements": ["ID badge", "keys", "coins", "pens"], "negative_prompt": "faded colors, poor lighting, background clutter", "diversity_tags": ["color similarity", "text region"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "eb19a4572a58", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.3846, "diversity_score": 0.4154, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d3_eb19a4572a58", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d3_eb19a4572a58/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25352516, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1503295, "qwen_elapsed_seconds": 5.18, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d3_eb19a4572a58/record.json"} +{"prompt_id": "clutter_distractor_d4_11a9bd230ad0", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic display with a black watch surrounded by similar-shaped gears and cogs, clear separation, sharp focus", "target_evidence": "black watch", "expected_visual_elements": ["black watch", "gears", "cogs"], "negative_prompt": "overlapping items, low contrast, poor lighting", "diversity_tags": ["shape similarity", "object count"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "11a9bd230ad0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic display with a", "max_similarity_to_other_prompt": 0.44, "diversity_score": 0.36, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_11a9bd230ad0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24720661, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6061195, "qwen_elapsed_seconds": 5.186, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_11a9bd230ad0/record.json"} +{"prompt_id": "clutter_distractor_d4_23e2e0468879", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic locker with a single blue pen amidst many black pens, clean boundaries, sharp focus, high contrast", "target_evidence": "single blue pen", "expected_visual_elements": ["locker", "many black pens"], "negative_prompt": "low contrast, overlapping objects, blurry edges", "diversity_tags": ["color similarity", "object scale variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "23e2e0468879", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker with a", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_23e2e0468879", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_23e2e0468879/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25589732, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 607559, "qwen_elapsed_seconds": 5.109, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_23e2e0468879/record.json"} +{"prompt_id": "clutter_distractor_d4_263c71fe3951", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic workbench with a single green key surrounded by dozens of black keys, high detail, clear foreground separation", "target_evidence": "single green key", "expected_visual_elements": ["workbench", "dozens of black keys"], "negative_prompt": "low contrast, blurry edges, overlapping objects", "diversity_tags": ["color similarity", "object scale variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "263c71fe3951", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_263c71fe3951", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_263c71fe3951/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27662153, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 741514, "qwen_elapsed_seconds": 5.124, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_263c71fe3951/record.json"} +{"prompt_id": "clutter_distractor_d4_31b9fe844977", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic market display with a single red pen surrounded by dozens of similar pens, clean boundaries, sharp focus", "target_evidence": "single red pen", "expected_visual_elements": ["market display", "dozens of similar pens"], "negative_prompt": "blurry objects, overlapping elements, low contrast, poor lighting", "diversity_tags": ["similar-color distractors", "high density"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "31b9fe844977", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_31b9fe844977", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_31b9fe844977/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27294904, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7959812, "qwen_elapsed_seconds": 5.235, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_31b9fe844977/record.json"} +{"prompt_id": "clutter_distractor_d4_43a4872c9b38", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic market display with a single red apple surrounded by 25 similar-looking oranges, clean boundaries, sharp focus, and high contrast", "target_evidence": "red apple", "expected_visual_elements": ["red apple", "oranges", "market display"], "negative_prompt": "blurry objects, overlapping elements, low contrast, poor lighting", "diversity_tags": ["color similarity", "high clutter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "43a4872c9b38", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.6, "diversity_score": 0.2, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_43a4872c9b38", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_43a4872c9b38/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25413465, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8586496, "qwen_elapsed_seconds": 5.103, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_43a4872c9b38/record.json"} +{"prompt_id": "clutter_distractor_d4_4444dce4f109", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic workbench with a single yellow key hidden among 35 similar keys, sharp focus, clear foreground, and high contrast", "target_evidence": "yellow key", "expected_visual_elements": ["yellow key", "keys", "workbench"], "negative_prompt": "low resolution, blurred details, overlapping elements", "diversity_tags": ["shape similarity", "clean boundaries"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4444dce4f109", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.4444, "diversity_score": 0.3556, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_4444dce4f109", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_4444dce4f109/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28370943, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4187677, "qwen_elapsed_seconds": 5.14, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_4444dce4f109/record.json"} +{"prompt_id": "clutter_distractor_d4_454b748fb55b", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "Synthetic photorealistic market display with a single red apple amidst dozens of green apples, clean boundaries, sharp focus on target", "target_evidence": "single red apple", "expected_visual_elements": ["red apple", "green apples"], "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", "diversity_tags": ["color contrast", "high density"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "454b748fb55b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_454b748fb55b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_454b748fb55b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25624714, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1774113, "qwen_elapsed_seconds": 5.148, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_454b748fb55b/record.json"} +{"prompt_id": "clutter_distractor_d4_5be1c3b7383a", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic locker filled with 30 identical blue pens, one unique green pen in focus, clean lines, and clear foreground separation", "target_evidence": "green pen", "expected_visual_elements": ["blue pens", "green pen", "locker"], "negative_prompt": "distracting textures, overlapping items, soft focus", "diversity_tags": ["shape similarity", "object scale variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5be1c3b7383a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker filled with", "max_similarity_to_other_prompt": 0.4138, "diversity_score": 0.3862, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_5be1c3b7383a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24595388, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6664834, "qwen_elapsed_seconds": 5.217, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_5be1c3b7383a/record.json"} +{"prompt_id": "clutter_distractor_d4_5bfbc80b56a2", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "Synthetic photorealistic shelf with a yellow label on top of many blue labels, distinct color, clear foreground separation", "target_evidence": "yellow label", "expected_visual_elements": ["yellow label", "blue labels"], "negative_prompt": "blurry, overlapping, low contrast", "diversity_tags": ["color contrast", "placement emphasis"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5bfbc80b56a2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.4583, "diversity_score": 0.3417, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_5bfbc80b56a2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26021712, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7542871, "qwen_elapsed_seconds": 5.296, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_5bfbc80b56a2/record.json"} +{"prompt_id": "clutter_distractor_d4_6c208faaaafa", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic shelf with a purple pen in a stack of similar-colored markers, clean boundaries, clear foreground", "target_evidence": "purple pen", "expected_visual_elements": ["purple pen", "similar-colored markers"], "negative_prompt": "overlapping items, low contrast, poor lighting", "diversity_tags": ["color similarity", "object count"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6c208faaaafa", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.4444, "diversity_score": 0.3556, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_6c208faaaafa", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_6c208faaaafa/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26661592, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9993344, "qwen_elapsed_seconds": 5.142, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_6c208faaaafa/record.json"} +{"prompt_id": "clutter_distractor_d4_71d67fddf097", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic market display with a single red balloon among hundreds of blue balloons, sharp focus, high contrast", "target_evidence": "single red balloon", "expected_visual_elements": ["market display", "hundreds of blue balloons"], "negative_prompt": "low contrast, overlapping objects, blurry edges", "diversity_tags": ["color similarity", "high distractor count"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "71d67fddf097", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.6522, "diversity_score": 0.1478, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_71d67fddf097", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_71d67fddf097/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24219131, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8726131, "qwen_elapsed_seconds": 5.201, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_71d67fddf097/record.json"} +{"prompt_id": "clutter_distractor_d4_7a819468a683", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic shelf with a single yellow key surrounded by dozens of identical keys, clear foreground separation, high detail", "target_evidence": "single yellow key", "expected_visual_elements": ["shelf", "dozens of identical keys"], "negative_prompt": "low resolution, overlapping items, low detail", "diversity_tags": ["shape similarity", "object scale variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7a819468a683", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_7a819468a683", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_7a819468a683/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25753108, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6946850, "qwen_elapsed_seconds": 5.233, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_7a819468a683/record.json"} +{"prompt_id": "clutter_distractor_d4_88f59b021a79", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic workbench with a single white ruler among dozens of wooden rulers, high detail, clear foreground separation", "target_evidence": "single white ruler", "expected_visual_elements": ["workbench", "dozens of wooden rulers"], "negative_prompt": "low detail, overlapping items, blurry edges", "diversity_tags": ["shape similarity", "object scale variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "88f59b021a79", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_88f59b021a79", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_88f59b021a79/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26649291, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1063384, "qwen_elapsed_seconds": 5.201, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_88f59b021a79/record.json"} +{"prompt_id": "clutter_distractor_d4_97fe5dcd2ae5", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic market display with a single white apple in a sea of red and green fruits, clean boundaries, clear focus", "target_evidence": "white apple", "expected_visual_elements": ["market display", "fruits"], "negative_prompt": "low contrast, blurred edges, overlapping elements, poor lighting", "diversity_tags": ["similar-color", "clear-boundaries"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "97fe5dcd2ae5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_97fe5dcd2ae5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23011897, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4374182, "qwen_elapsed_seconds": 5.247, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_97fe5dcd2ae5/record.json"} +{"prompt_id": "clutter_distractor_d4_9b4e5450a8e1", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic market display with a single red apple among hundreds of green apples, clean boundaries, sharp focus, high contrast", "target_evidence": "single red apple", "expected_visual_elements": ["market display", "hundreds of green apples"], "negative_prompt": "low contrast, blurry edges, overlapping objects", "diversity_tags": ["color similarity", "high distractor count"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "9b4e5450a8e1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_9b4e5450a8e1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24204502, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8857956, "qwen_elapsed_seconds": 5.178, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_9b4e5450a8e1/record.json"} +{"prompt_id": "clutter_distractor_d4_a7a966f3fcee", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic bulletin board with a single blue sticker among many similar stickers, clear foreground separation, high contrast", "target_evidence": "blue sticker", "expected_visual_elements": ["bulletin board", "stickers"], "negative_prompt": "low resolution, out-of-focus, overlapping elements, poor lighting", "diversity_tags": ["similar-shape", "clear-boundaries"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a7a966f3fcee", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_a7a966f3fcee", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20326051, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6521731, "qwen_elapsed_seconds": 5.223, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_a7a966f3fcee/record.json"} +{"prompt_id": "clutter_distractor_d4_c6f250be29ba", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic bulletin board with a green folder surrounded by sticky notes and rulers, high contrast, clear foreground separation", "target_evidence": "green folder", "expected_visual_elements": ["green folder", "sticky notes", "rulers"], "negative_prompt": "faded colors, poor lighting, merged objects", "diversity_tags": ["shape similarity", "layout variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c6f250be29ba", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.4643, "diversity_score": 0.3357, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_c6f250be29ba", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_c6f250be29ba/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24048938, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2086878, "qwen_elapsed_seconds": 5.207, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_c6f250be29ba/record.json"} +{"prompt_id": "clutter_distractor_d4_ce19ddf16fef", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic bulletin board with a single green pencil surrounded by dozens of black pencils, high contrast, sharp focus", "target_evidence": "single green pencil", "expected_visual_elements": ["bulletin board", "dozens of black pencils"], "negative_prompt": "low contrast, overlapping objects, low detail", "diversity_tags": ["color similarity", "object scale variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ce19ddf16fef", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_ce19ddf16fef", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29519981, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2624438, "qwen_elapsed_seconds": 5.184, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_ce19ddf16fef/record.json"} +{"prompt_id": "clutter_distractor_d4_ea4e61598d94", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "Synthetic photorealistic bulletin board with a white label on top of many colored labels, sharp focus, distinct placement", "target_evidence": "white label", "expected_visual_elements": ["white label", "colored labels"], "negative_prompt": "overlapping, low resolution, poor lighting", "diversity_tags": ["color contrast", "placement emphasis"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ea4e61598d94", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.44, "diversity_score": 0.36, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_ea4e61598d94", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_ea4e61598d94/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22339680, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1371864, "qwen_elapsed_seconds": 5.222, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_ea4e61598d94/record.json"} +{"prompt_id": "clutter_distractor_d4_ff2b8c5d1325", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 4, "image_prompt": "synthetic photorealistic workbench with a blue key in a pile of screws and nails, distinct edges, clean boundaries", "target_evidence": "blue key", "expected_visual_elements": ["blue key", "screws", "nails"], "negative_prompt": "overlapping items, low resolution, poor focus", "diversity_tags": ["object scale", "material variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ff2b8c5d1325", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.3704, "diversity_score": 0.4296, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d4_ff2b8c5d1325", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23168712, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9683140, "qwen_elapsed_seconds": 5.24, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d4_ff2b8c5d1325/record.json"} +{"prompt_id": "clutter_distractor_d5_04829301b77b", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "synthetic photorealistic market display with a red wrench among scattered tools, sharp focus on target, clean boundaries, high contrast lighting", "target_evidence": "red wrench", "expected_visual_elements": ["red wrench", "scattered tools", "market display"], "negative_prompt": "blurry objects, low contrast, overlapping elements, poor lighting", "diversity_tags": ["color similarity", "tool clutter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "04829301b77b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.5833, "diversity_score": 0.2167, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_04829301b77b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_04829301b77b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25227361, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4186258, "qwen_elapsed_seconds": 5.215, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_04829301b77b/record.json"} +{"prompt_id": "clutter_distractor_d5_06238c20c828", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "Synthetic photorealistic tabletop with a single white cup among 40 similar-shaped ceramic objects, high contrast and sharp focus", "target_evidence": "white cup", "expected_visual_elements": ["ceramic objects", "white cup"], "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", "diversity_tags": ["similar-shape distractors", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "06238c20c828", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tabletop with a", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_06238c20c828", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_06238c20c828/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29244267, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9106413, "qwen_elapsed_seconds": 5.22, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_06238c20c828/record.json"} +{"prompt_id": "clutter_distractor_d5_1aa08434a01a", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "synthetic photorealistic shelf with a red pen surrounded by similar-shaped pens, sharp focus on target, clear foreground", "target_evidence": "red pen", "expected_visual_elements": ["red pen", "similar-shaped pens", "shelf"], "negative_prompt": "overlapping pens, low contrast, poor focus, blurry background", "diversity_tags": ["shape similarity", "office items"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1aa08434a01a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.6087, "diversity_score": 0.1913, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_1aa08434a01a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_1aa08434a01a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26731752, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1239610, "qwen_elapsed_seconds": 5.233, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_1aa08434a01a/record.json"} +{"prompt_id": "clutter_distractor_d5_24ee9c8ef6d0", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "synthetic photorealistic market display with a single red keychain among 45 plastic bottle caps, clean boundaries, sharp focus, high contrast", "target_evidence": "red keychain", "expected_visual_elements": ["plastic bottle caps", "red keychain"], "negative_prompt": "low contrast, blurry edges, overlapping objects, poor lighting", "diversity_tags": ["similar-color distractors", "high distractor count"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "24ee9c8ef6d0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.6, "diversity_score": 0.2, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_24ee9c8ef6d0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20598557, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7799995, "qwen_elapsed_seconds": 5.222, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_24ee9c8ef6d0/record.json"} +{"prompt_id": "clutter_distractor_d5_2e2224605b5c", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "synthetic photorealistic bulletin board with a yellow label amidst similar-color labels, clean foreground, sharp focus on target", "target_evidence": "yellow label", "expected_visual_elements": ["yellow label", "similar-color labels", "bulletin board"], "negative_prompt": "overlapping labels, low contrast, poor focus, blurry background", "diversity_tags": ["color similarity", "label clutter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2e2224605b5c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.44, "diversity_score": 0.36, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_2e2224605b5c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_2e2224605b5c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29785390, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1684863, "qwen_elapsed_seconds": 5.168, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_2e2224605b5c/record.json"} +{"prompt_id": "clutter_distractor_d5_3979ce7f86ea", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "synthetic photorealistic shelf with a single blue USB drive among 50 white paper clips, high contrast, clean edges, visible target", "target_evidence": "blue USB drive", "expected_visual_elements": ["white paper clips", "blue USB drive"], "negative_prompt": "low contrast, overlapping items, poor lighting, low detail", "diversity_tags": ["similar-color distractors", "high distractor count"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "3979ce7f86ea", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.3333, "diversity_score": 0.4667, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_3979ce7f86ea", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29446383, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7962205, "qwen_elapsed_seconds": 5.256, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_3979ce7f86ea/record.json"} +{"prompt_id": "clutter_distractor_d5_62c3301efb9c", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "Synthetic photorealistic display with a single orange box visible among 40 similar-sized cardboard boxes, clean boundaries and high contrast", "target_evidence": "orange box", "expected_visual_elements": ["cardboard boxes", "orange box"], "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", "diversity_tags": ["similar-scale distractors", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "62c3301efb9c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic display with a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_62c3301efb9c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_62c3301efb9c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22543368, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8675430, "qwen_elapsed_seconds": 5.192, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_62c3301efb9c/record.json"} +{"prompt_id": "clutter_distractor_d5_637cd118a043", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "synthetic photorealistic workbench with a single yellow padlock among 38 metal screws, clear foreground separation, sharp focus, distinct textures", "target_evidence": "yellow padlock", "expected_visual_elements": ["metal screws", "yellow padlock"], "negative_prompt": "low resolution, blurred focus, cluttered background, overlapping objects", "diversity_tags": ["similar-shape distractors", "distinct textures"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "637cd118a043", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.4444, "diversity_score": 0.3556, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_637cd118a043", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_637cd118a043/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26491783, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7792357, "qwen_elapsed_seconds": 5.175, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_637cd118a043/record.json"} +{"prompt_id": "clutter_distractor_d5_79441f6c0574", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "Synthetic photorealistic workbench with a single green bottle among 40 similar-shaped glass containers, high contrast and clean boundaries", "target_evidence": "green bottle", "expected_visual_elements": ["glass containers", "green bottle"], "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", "diversity_tags": ["similar-shape distractors", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "79441f6c0574", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.56, "diversity_score": 0.24, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_79441f6c0574", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_79441f6c0574/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26333002, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6156372, "qwen_elapsed_seconds": 5.158, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_79441f6c0574/record.json"} +{"prompt_id": "clutter_distractor_d5_82234aef50b3", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "Synthetic photorealistic bulletin board with a blue label standing out among 40 similar-colored paper notes, clear foreground separation", "target_evidence": "blue label", "expected_visual_elements": ["paper notes", "blue label"], "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", "diversity_tags": ["similar-color distractors", "clear foreground"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "82234aef50b3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_82234aef50b3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_82234aef50b3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28422949, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5478104, "qwen_elapsed_seconds": 5.195, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_82234aef50b3/record.json"} +{"prompt_id": "clutter_distractor_d5_85f68bfdbfa5", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "synthetic photorealistic locker filled with a single orange ruler among 37 small tools, clear foreground separation, high contrast, visible target", "target_evidence": "orange ruler", "expected_visual_elements": ["small tools", "orange ruler"], "negative_prompt": "low contrast, blurry edges, overlapping items, poor lighting", "diversity_tags": ["similar-shape distractors", "high distractor count"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "85f68bfdbfa5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker filled with", "max_similarity_to_other_prompt": 0.4074, "diversity_score": 0.3926, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_85f68bfdbfa5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23085812, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6095068, "qwen_elapsed_seconds": 5.139, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_85f68bfdbfa5/record.json"} +{"prompt_id": "clutter_distractor_d5_9bc239a79071", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "synthetic photorealistic locker filled with multiple small tools, a yellow wrench stands out with clean edges and sharp focus", "target_evidence": "yellow wrench", "expected_visual_elements": ["tools", "locker"], "negative_prompt": "low detail, blurred edges, overlapping objects", "diversity_tags": ["scale variation", "color similarity"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "9bc239a79071", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker filled with", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_9bc239a79071", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_9bc239a79071/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23987141, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2830834, "qwen_elapsed_seconds": 5.138, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_9bc239a79071/record.json"} +{"prompt_id": "clutter_distractor_d5_a3a3c294f30d", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "synthetic photorealistic bulletin board with a single green notebook among 42 sticky notes, sharp focus, clean boundaries, distinct colors", "target_evidence": "green notebook", "expected_visual_elements": ["sticky notes", "green notebook"], "negative_prompt": "blurred objects, overlapping elements, poor lighting, low detail", "diversity_tags": ["similar-shape distractors", "distinct colors"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a3a3c294f30d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic bulletin board with", "max_similarity_to_other_prompt": 0.4, "diversity_score": 0.4, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_a3a3c294f30d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21449348, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3614937, "qwen_elapsed_seconds": 5.171, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_a3a3c294f30d/record.json"} +{"prompt_id": "clutter_distractor_d5_a4819ff5cb89", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "Synthetic photorealistic locker filled with 40 small screws, one large silver screw stands out with sharp focus and clean edges", "target_evidence": "large silver screw", "expected_visual_elements": ["small screws", "large silver screw"], "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", "diversity_tags": ["similar-scale distractors", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a4819ff5cb89", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic locker filled with", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_a4819ff5cb89", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27461534, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7471570, "qwen_elapsed_seconds": 5.167, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_a4819ff5cb89/record.json"} +{"prompt_id": "clutter_distractor_d5_a675e91f33b7", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "Synthetic photorealistic workbench with a single blue screw visible among 40 similar-shaped metallic items, high contrast and clear boundaries", "target_evidence": "blue screw", "expected_visual_elements": ["metallic items", "blue screw"], "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", "diversity_tags": ["similar-shape distractors", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a675e91f33b7", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.56, "diversity_score": 0.24, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_a675e91f33b7", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_a675e91f33b7/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20985734, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7346072, "qwen_elapsed_seconds": 5.16, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_a675e91f33b7/record.json"} +{"prompt_id": "clutter_distractor_d5_a722baad73c5", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "synthetic photorealistic market display with a red rubber duck among dozens of colorful balloons, clean boundaries, sharp focus on duck", "target_evidence": "red rubber duck", "expected_visual_elements": ["balloons", "market display"], "negative_prompt": "blurry focus, low contrast, overlapping objects", "diversity_tags": ["color similarity", "object scale variation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a722baad73c5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.56, "diversity_score": 0.24, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_a722baad73c5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_a722baad73c5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29163573, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6938740, "qwen_elapsed_seconds": 5.184, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_a722baad73c5/record.json"} +{"prompt_id": "clutter_distractor_d5_b0211aa536ef", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "Synthetic photorealistic market display with a single red key visible among 40 similar-shaped plastic items, clean boundaries, sharp focus", "target_evidence": "single red key", "expected_visual_elements": ["plastic items", "red key"], "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", "diversity_tags": ["similar-shape distractors", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b0211aa536ef", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic market display with", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_b0211aa536ef", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_b0211aa536ef/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29478110, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1585783, "qwen_elapsed_seconds": 5.114, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_b0211aa536ef/record.json"} +{"prompt_id": "clutter_distractor_d5_d1ee86278851", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "Synthetic photorealistic shelf with a single green book standing out among 40 similar-sized paperbacks, sharp focus and clean edges", "target_evidence": "green book", "expected_visual_elements": ["paperbacks", "green book"], "negative_prompt": "low contrast, blurry objects, overlapping elements, poor lighting", "diversity_tags": ["similar-scale distractors", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d1ee86278851", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.3929, "diversity_score": 0.4071, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_d1ee86278851", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_d1ee86278851/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27530094, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6748699, "qwen_elapsed_seconds": 5.126, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_d1ee86278851/record.json"} +{"prompt_id": "clutter_distractor_d5_dbb61cceb7f8", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "synthetic photorealistic shelf with a blue key surrounded by similar-shaped keys, clean foreground separation, sharp focus on target", "target_evidence": "blue key", "expected_visual_elements": ["blue key", "similar-shaped keys", "shelf"], "negative_prompt": "overlapping keys, low contrast, poor focus, blurry background", "diversity_tags": ["shape similarity", "key clutter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "dbb61cceb7f8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shelf with a", "max_similarity_to_other_prompt": 0.6087, "diversity_score": 0.1913, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_dbb61cceb7f8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24433138, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3994459, "qwen_elapsed_seconds": 5.111, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_dbb61cceb7f8/record.json"} +{"prompt_id": "clutter_distractor_d5_f953bafa1186", "run_name": "flux2_qil_decomposition_pilot", "category": "clutter_distractor", "difficulty": 5, "image_prompt": "synthetic photorealistic workbench with a green pen among many pens, sharp focus on target, clean boundaries, high contrast", "target_evidence": "green pen", "expected_visual_elements": ["green pen", "many pens", "workbench"], "negative_prompt": "overlapping pens, low contrast, poor focus, blurry background", "diversity_tags": ["color similarity", "office items"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f953bafa1186", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic workbench with a", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "clutter_distractor_d5_f953bafa1186", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/clutter_distractor/clutter_distractor_d5_f953bafa1186/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28371434, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_00.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_01.png", "layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6821332, "qwen_elapsed_seconds": 5.155, "qwen_record_path": "layers-x/k3/clutter_distractor/clutter_distractor_d5_f953bafa1186/record.json"} +{"prompt_id": "multi_object_relations_d1_018f62776b1d", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "Synthetic photorealistic scene with a mannequin holding a key, clean boundaries and sharp focus on the key and hand, relation clearly visible", "target_evidence": "mannequin holding key", "expected_visual_elements": ["mannequin", "key"], "negative_prompt": "blurry, low contrast, overlapping objects", "diversity_tags": ["holding", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "018f62776b1d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.75, "diversity_score": 0.05, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_018f62776b1d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_018f62776b1d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26286557, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5356689, "qwen_elapsed_seconds": 5.173, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/record.json"} +{"prompt_id": "multi_object_relations_d1_2056f61d59a6", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "Synthetic photorealistic view of a mannequin standing behind a desk, sharp focus and clear spatial relation between mannequin and desk", "target_evidence": "mannequin behind desk", "expected_visual_elements": ["mannequin", "desk"], "negative_prompt": "blurry, low depth, overlapping elements", "diversity_tags": ["behind", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2056f61d59a6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4545, "diversity_score": 0.3455, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_2056f61d59a6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_2056f61d59a6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21921962, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8068497, "qwen_elapsed_seconds": 5.156, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/record.json"} +{"prompt_id": "multi_object_relations_d1_26310eaeece0", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "synthetic photorealistic view of a bag attached to a backpack strap, clear foreground and sharp focus on the attachment point", "target_evidence": "bag attached to backpack strap", "expected_visual_elements": ["bag", "backpack strap"], "negative_prompt": "blurry, low contrast, watermarks, real people, faces", "diversity_tags": ["attached to", "backpack"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "26310eaeece0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4167, "diversity_score": 0.3833, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_26310eaeece0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_26310eaeece0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27538671, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4635157, "qwen_elapsed_seconds": 5.209, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/record.json"} +{"prompt_id": "multi_object_relations_d1_28810914b7b6", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "Synthetic photorealistic scene with a mannequin placing a bag on a table, clean boundaries and clear relation between mannequin and bag", "target_evidence": "mannequin placing bag on table", "expected_visual_elements": ["mannequin", "bag", "table"], "negative_prompt": "low detail, out of focus, unclear positioning", "diversity_tags": ["on top of", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "28810914b7b6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.5238, "diversity_score": 0.2762, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_28810914b7b6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_28810914b7b6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28738385, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 587097, "qwen_elapsed_seconds": 5.172, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/record.json"} +{"prompt_id": "multi_object_relations_d1_2c93806efe70", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "Synthetic photorealistic setup with a tool on top of a bag, high contrast and clear visual relation between tool and bag", "target_evidence": "tool on top of bag", "expected_visual_elements": ["tool", "bag"], "negative_prompt": "low resolution, unclear positioning, overlapping objects", "diversity_tags": ["on top of", "bag"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2c93806efe70", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.6364, "diversity_score": 0.1636, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_2c93806efe70", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_2c93806efe70/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 30003506, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7539450, "qwen_elapsed_seconds": 5.155, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/record.json"} +{"prompt_id": "multi_object_relations_d1_431f8a0e9b3d", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "synthetic photorealistic scene with a mannequin holding a toolbox on a desk, clean boundaries and sharp focus for clear held-by relation", "target_evidence": "mannequin holding toolbox", "expected_visual_elements": ["mannequin", "toolbox", "desk"], "negative_prompt": "blurry, low contrast, overlapping objects, real people, faces", "diversity_tags": ["holding", "mannequin", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "431f8a0e9b3d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.6957, "diversity_score": 0.1043, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_431f8a0e9b3d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 30165319, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 405868, "qwen_elapsed_seconds": 5.145, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/record.json"} +{"prompt_id": "multi_object_relations_d1_4d1095976c42", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "Synthetic photorealistic setup with a box on top of a shelf, high contrast between box and shelf, relation visually legible", "target_evidence": "box on shelf", "expected_visual_elements": ["box", "shelf"], "negative_prompt": "low resolution, out of focus, cluttered background", "diversity_tags": ["on top of", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4d1095976c42", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_4d1095976c42", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_4d1095976c42/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25329557, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1971817, "qwen_elapsed_seconds": 5.207, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/record.json"} +{"prompt_id": "multi_object_relations_d1_4d6b6f9003fb", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "Synthetic photorealistic image of a bag inside a container, clear foreground separation and sharp edges between bag and container", "target_evidence": "bag inside container", "expected_visual_elements": ["bag", "container"], "negative_prompt": "foggy, washed out colors, objects partially obscured", "diversity_tags": ["inside", "container"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4d6b6f9003fb", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5789, "diversity_score": 0.2211, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_4d6b6f9003fb", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20849906, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5595332, "qwen_elapsed_seconds": 5.273, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/record.json"} +{"prompt_id": "multi_object_relations_d1_528663b80801", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "Synthetic photorealistic scene with a tool attached to a sign, clean lines and clear relation between tool and sign", "target_evidence": "tool attached to sign", "expected_visual_elements": ["tool", "sign"], "negative_prompt": "distorted shapes, unclear connections, low detail", "diversity_tags": ["attached-to", "sign"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "528663b80801", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.65, "diversity_score": 0.15, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_528663b80801", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_528663b80801/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22459641, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1971405, "qwen_elapsed_seconds": 5.085, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/record.json"} +{"prompt_id": "multi_object_relations_d1_52acc1c2fbb9", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "synthetic photorealistic setup with a tool under a container, clean separation and high contrast between the tool and container base", "target_evidence": "tool under container", "expected_visual_elements": ["tool", "container"], "negative_prompt": "blurry, low contrast, watermarks, real people, faces", "diversity_tags": ["under", "container"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "52acc1c2fbb9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.6316, "diversity_score": 0.1684, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_52acc1c2fbb9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28233303, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9946804, "qwen_elapsed_seconds": 5.154, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/record.json"} +{"prompt_id": "multi_object_relations_d1_586bda1348a2", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "Synthetic photorealistic scene with a mannequin holding a sign, clear foreground separation and sharp focus on the sign and hand", "target_evidence": "mannequin holding sign", "expected_visual_elements": ["mannequin", "sign"], "negative_prompt": "blurry, low contrast, overlapping objects", "diversity_tags": ["holding", "sign"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "586bda1348a2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.8, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_586bda1348a2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_586bda1348a2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20335950, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9694648, "qwen_elapsed_seconds": 5.092, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/record.json"} +{"prompt_id": "multi_object_relations_d1_6b46aeee2b20", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "synthetic photorealistic image of a mannequin with a tool attached to its arm, clear boundaries for attached-to relation", "target_evidence": "tool attached to mannequin arm", "expected_visual_elements": ["mannequin", "tool", "arm"], "negative_prompt": "overlapping, low detail, real people, faces", "diversity_tags": ["attached-to", "mannequin", "tool"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6b46aeee2b20", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5909, "diversity_score": 0.2091, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_6b46aeee2b20", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29391804, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5613533, "qwen_elapsed_seconds": 5.168, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/record.json"} +{"prompt_id": "multi_object_relations_d1_80c1fe8f92a4", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "synthetic photorealistic image of a tool inside a toolbox, clean boundaries and high contrast between the tool and box interior", "target_evidence": "tool inside toolbox", "expected_visual_elements": ["tool", "toolbox"], "negative_prompt": "blurry, low contrast, watermarks, real people, faces", "diversity_tags": ["inside", "toolbox"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "80c1fe8f92a4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5238, "diversity_score": 0.2762, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_80c1fe8f92a4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20649079, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4576292, "qwen_elapsed_seconds": 5.139, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/record.json"} +{"prompt_id": "multi_object_relations_d1_98504a155dad", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "synthetic photorealistic setup with a bag placed on top of a shelf, high contrast for visible on-top relation", "target_evidence": "bag on top of shelf", "expected_visual_elements": ["bag", "shelf", "shelf unit"], "negative_prompt": "overlapping, low detail, real people, faces", "diversity_tags": ["on-top", "shelf", "bag"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "98504a155dad", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.7222, "diversity_score": 0.0778, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_98504a155dad", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_98504a155dad/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21842447, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7329201, "qwen_elapsed_seconds": 5.181, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/record.json"} +{"prompt_id": "multi_object_relations_d1_9ab06ddd4ac2", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "synthetic photorealistic setup with a box on top of a shelf, high contrast and clean separation between the box and shelf edges", "target_evidence": "box on top of shelf", "expected_visual_elements": ["box", "shelf"], "negative_prompt": "blurry, low contrast, watermarks, real people, faces", "diversity_tags": ["on top of", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "9ab06ddd4ac2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.8, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_9ab06ddd4ac2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28600316, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7036460, "qwen_elapsed_seconds": 5.196, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/record.json"} +{"prompt_id": "multi_object_relations_d1_a5c5b8ab9db1", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "Synthetic photorealistic image of a key inside a box, sharp focus and clear visual separation between key and box", "target_evidence": "key inside box", "expected_visual_elements": ["key", "box"], "negative_prompt": "foggy, washed out colors, overlapping objects", "diversity_tags": ["inside", "box"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a5c5b8ab9db1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5789, "diversity_score": 0.2211, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_a5c5b8ab9db1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24101894, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5309862, "qwen_elapsed_seconds": 5.196, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/record.json"} +{"prompt_id": "multi_object_relations_d1_af1dae7eec23", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "synthetic photorealistic scene with a sign behind a desk, sharp focus and clear separation between the sign and desk surface", "target_evidence": "sign behind desk", "expected_visual_elements": ["sign", "desk"], "negative_prompt": "blurry, low contrast, watermarks, real people, faces", "diversity_tags": ["behind", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "af1dae7eec23", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.5714, "diversity_score": 0.2286, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_af1dae7eec23", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_af1dae7eec23/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26233414, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4368367, "qwen_elapsed_seconds": 5.233, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/record.json"} +{"prompt_id": "multi_object_relations_d1_b8bb45add71b", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "synthetic photorealistic scene with a mannequin holding a key, clean boundaries and sharp focus on the key and hand, clear foreground separation", "target_evidence": "mannequin holding key", "expected_visual_elements": ["mannequin", "key"], "negative_prompt": "blurry, low contrast, watermarks, real people, faces", "diversity_tags": ["holding", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b8bb45add71b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.8, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_b8bb45add71b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_b8bb45add71b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28969010, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6098582, "qwen_elapsed_seconds": 5.228, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/record.json"} +{"prompt_id": "multi_object_relations_d1_dd45a37dffac", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "Synthetic photorealistic setup with a tool under a container, high contrast and clear spatial relation between tool and container", "target_evidence": "tool under container", "expected_visual_elements": ["tool", "container"], "negative_prompt": "low contrast, ambiguous positioning, overlapping objects", "diversity_tags": ["under", "container"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "dd45a37dffac", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.6316, "diversity_score": 0.1684, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_dd45a37dffac", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_dd45a37dffac/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20420612, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7009906, "qwen_elapsed_seconds": 5.143, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/record.json"} +{"prompt_id": "multi_object_relations_d1_fb7d932b5707", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 1, "image_prompt": "synthetic photorealistic scene with a container on top of a shelf, high contrast for visible on-top relation", "target_evidence": "container on top of shelf", "expected_visual_elements": ["container", "shelf", "shelf unit"], "negative_prompt": "blurry, low detail, real people, faces", "diversity_tags": ["on-top", "shelf", "container"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "fb7d932b5707", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.7222, "diversity_score": 0.0778, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d1_fb7d932b5707", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_fb7d932b5707/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21246195, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4532813, "qwen_elapsed_seconds": 5.164, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/record.json"} +{"prompt_id": "multi_object_relations_d2_0a361d43b600", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic scene with a mannequin holding a bag, clean boundaries and sharp focus on the hand and bag", "target_evidence": "mannequin holding bag", "expected_visual_elements": ["mannequin", "bag"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["holding", "bag"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0a361d43b600", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.8333, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_0a361d43b600", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_0a361d43b600/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21501073, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8016721, "qwen_elapsed_seconds": 5.107, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/record.json"} +{"prompt_id": "multi_object_relations_d2_0c527d58a238", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic scene with a sign attached to a wall, clear foreground separation and high contrast for visual inspection", "target_evidence": "sign attached to wall", "expected_visual_elements": ["sign", "wall", "clear foreground"], "negative_prompt": "overlapping, low detail, unrealistic lighting, blurry edges", "diversity_tags": ["attached to", "sign", "wall"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0c527d58a238", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.75, "diversity_score": 0.05, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_0c527d58a238", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_0c527d58a238/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21172209, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9882009, "qwen_elapsed_seconds": 5.223, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/record.json"} +{"prompt_id": "multi_object_relations_d2_21de220e975c", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic image of a tool inside a bag, clean boundaries and sharp focus on the tool and bag", "target_evidence": "tool inside bag", "expected_visual_elements": ["tool", "bag"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["inside", "bag"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "21de220e975c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.875, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_21de220e975c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_21de220e975c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25827693, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 707075, "qwen_elapsed_seconds": 5.2, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/record.json"} +{"prompt_id": "multi_object_relations_d2_2c78cb9367b9", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic image of a bag inside a container, clean boundaries and sharp focus on the bag and container", "target_evidence": "bag inside container", "expected_visual_elements": ["bag", "container"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["inside", "container"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2c78cb9367b9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.875, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_2c78cb9367b9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23681224, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6345146, "qwen_elapsed_seconds": 5.163, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/record.json"} +{"prompt_id": "multi_object_relations_d2_31914a9d70f8", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic setup with a box on top of a desk, clear foreground separation and high contrast between the box and desk", "target_evidence": "box on top of desk", "expected_visual_elements": ["box", "desk"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["on top of", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "31914a9d70f8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.8947, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_31914a9d70f8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_31914a9d70f8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23431206, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 455263, "qwen_elapsed_seconds": 5.138, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/record.json"} +{"prompt_id": "multi_object_relations_d2_40d6a9b7e720", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic image of a container with a key inside, clean boundaries and sharp focus for relation clarity", "target_evidence": "key inside container", "expected_visual_elements": ["container", "key", "sharp focus"], "negative_prompt": "blurry, overlapping, low contrast, unrealistic lighting", "diversity_tags": ["inside", "container", "key"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "40d6a9b7e720", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.6, "diversity_score": 0.2, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_40d6a9b7e720", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27511360, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6786192, "qwen_elapsed_seconds": 5.134, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/record.json"} +{"prompt_id": "multi_object_relations_d2_412c56ac4795", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic setup with a bag on top of a shelf, high contrast and clear foreground separation for visual inspection", "target_evidence": "bag on top of shelf", "expected_visual_elements": ["bag", "shelf", "clear foreground"], "negative_prompt": "overlapping, low detail, unrealistic lighting, blurry edges", "diversity_tags": ["on top of", "shelf", "bag"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "412c56ac4795", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.8095, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_412c56ac4795", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_412c56ac4795/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24115452, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 414113, "qwen_elapsed_seconds": 5.127, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/record.json"} +{"prompt_id": "multi_object_relations_d2_4db6b2035025", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic image of a tool attached to a shelf, clear foreground separation and high contrast between the tool and shelf", "target_evidence": "tool attached to shelf", "expected_visual_elements": ["tool", "shelf"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["attached to", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4db6b2035025", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5909, "diversity_score": 0.2091, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_4db6b2035025", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_4db6b2035025/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20963077, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 963109, "qwen_elapsed_seconds": 5.107, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/record.json"} +{"prompt_id": "multi_object_relations_d2_5126e8e36b8d", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic scene with a mannequin holding a wrench, clean boundaries and sharp focus on the tool and hand", "target_evidence": "mannequin holding wrench", "expected_visual_elements": ["mannequin", "wrench"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["holding", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5126e8e36b8d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.8333, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_5126e8e36b8d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20810882, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 450326, "qwen_elapsed_seconds": 5.199, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/record.json"} +{"prompt_id": "multi_object_relations_d2_587f3e361075", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic scene with a box behind a chair, clean boundaries and sharp focus for clear relation inspection", "target_evidence": "box behind chair", "expected_visual_elements": ["box", "chair", "sharp focus"], "negative_prompt": "blurry, overlapping, low contrast, unrealistic lighting", "diversity_tags": ["behind", "chair", "box"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "587f3e361075", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.7143, "diversity_score": 0.0857, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_587f3e361075", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_587f3e361075/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27876213, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3835553, "qwen_elapsed_seconds": 5.224, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/record.json"} +{"prompt_id": "multi_object_relations_d2_639e608a1f85", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic setup with a box on top of a shelf, clear foreground separation and high contrast between the box and shelf", "target_evidence": "box on top of shelf", "expected_visual_elements": ["box", "shelf"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["on top of", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "639e608a1f85", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.8947, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_639e608a1f85", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_639e608a1f85/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24140608, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4550962, "qwen_elapsed_seconds": 5.176, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/record.json"} +{"prompt_id": "multi_object_relations_d2_6ab23ebd01c6", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic image of a mannequin holding a cup, clean boundaries and sharp focus for clear relation inspection", "target_evidence": "mannequin holding cup", "expected_visual_elements": ["mannequin", "cup", "sharp focus"], "negative_prompt": "blurry, overlapping, low contrast, unrealistic lighting", "diversity_tags": ["holding", "mannequin", "cup"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6ab23ebd01c6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.6364, "diversity_score": 0.1636, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_6ab23ebd01c6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22240429, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9874851, "qwen_elapsed_seconds": 5.129, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/record.json"} +{"prompt_id": "multi_object_relations_d2_6b2da3ae773a", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic scene with a bag inside a box, clear foreground separation and high contrast for visual inspection", "target_evidence": "bag inside box", "expected_visual_elements": ["bag", "box", "clear foreground"], "negative_prompt": "overlapping, low detail, unrealistic lighting, blurry edges", "diversity_tags": ["inside", "box", "bag"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6b2da3ae773a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_6b2da3ae773a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 30250693, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9963444, "qwen_elapsed_seconds": 5.202, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/record.json"} +{"prompt_id": "multi_object_relations_d2_6bbeb1d0bad7", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic setup with a tool under a table, clean boundaries and sharp focus on the tool and table", "target_evidence": "tool under table", "expected_visual_elements": ["tool", "table"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["under", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6bbeb1d0bad7", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.619, "diversity_score": 0.181, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_6bbeb1d0bad7", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26374483, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3150863, "qwen_elapsed_seconds": 5.181, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/record.json"} +{"prompt_id": "multi_object_relations_d2_81fb8f2bf58d", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic scene with a mannequin holding a sign, clear foreground separation and high contrast between the hand and sign", "target_evidence": "mannequin holding sign", "expected_visual_elements": ["mannequin", "sign"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["holding", "sign"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "81fb8f2bf58d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.7895, "diversity_score": 0.0105, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_81fb8f2bf58d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20501665, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4851240, "qwen_elapsed_seconds": 5.19, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/record.json"} +{"prompt_id": "multi_object_relations_d2_add4beb155f4", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic image of a mannequin with a hat on its head, high contrast and clear foreground separation for visual inspection", "target_evidence": "hat on mannequin head", "expected_visual_elements": ["mannequin", "hat", "clear foreground"], "negative_prompt": "overlapping, low detail, unrealistic lighting, blurry edges", "diversity_tags": ["on top of", "mannequin", "hat"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "add4beb155f4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.175, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_add4beb155f4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_add4beb155f4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29113177, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2090035, "qwen_elapsed_seconds": 5.179, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/record.json"} +{"prompt_id": "multi_object_relations_d2_b1a4e86c9249", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic scene with a sign behind a mannequin, clear foreground separation and high contrast between the sign and figure", "target_evidence": "sign behind mannequin", "expected_visual_elements": ["sign", "mannequin"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["behind", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b1a4e86c9249", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.7895, "diversity_score": 0.0105, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_b1a4e86c9249", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24435050, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1438495, "qwen_elapsed_seconds": 5.222, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/record.json"} +{"prompt_id": "multi_object_relations_d2_dd6c090b67d8", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic setup with a book under a lamp, clean boundaries and sharp focus for clear relation inspection", "target_evidence": "book under lamp", "expected_visual_elements": ["book", "lamp", "sharp focus"], "negative_prompt": "blurry, overlapping, low contrast, unrealistic lighting", "diversity_tags": ["under", "lamp", "book"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "dd6c090b67d8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.619, "diversity_score": 0.181, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_dd6c090b67d8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24632545, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4669038, "qwen_elapsed_seconds": 5.189, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/record.json"} +{"prompt_id": "multi_object_relations_d2_e10e5f2a5b78", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic scene with a mannequin holding a wrench above a toolbox, clean boundaries and sharp focus for clear relation inspection", "target_evidence": "mannequin holding wrench", "expected_visual_elements": ["mannequin", "wrench", "toolbox"], "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", "diversity_tags": ["holding", "tools", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e10e5f2a5b78", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_e10e5f2a5b78", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20484070, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1626980, "qwen_elapsed_seconds": 5.168, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/record.json"} +{"prompt_id": "multi_object_relations_d2_f48653185696", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 2, "image_prompt": "synthetic photorealistic setup with a tool on top of a desk, high contrast and clear foreground separation for visual inspection", "target_evidence": "tool on desk", "expected_visual_elements": ["tool", "desk", "clear foreground"], "negative_prompt": "overlapping, low detail, unrealistic lighting, blurry edges", "diversity_tags": ["on top of", "desk", "tool"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f48653185696", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.8095, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d2_f48653185696", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_f48653185696/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29235481, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5328890, "qwen_elapsed_seconds": 5.153, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/record.json"} +{"prompt_id": "multi_object_relations_d3_074db631b159", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "Synthetic photorealistic setup with a bag on a desk, a pen inside the bag, and a sign next to the desk, high contrast and clear separation.", "target_evidence": "pen inside bag", "expected_visual_elements": ["bag", "pen", "desk", "sign"], "negative_prompt": "low resolution, unclear boundaries, cluttered background", "diversity_tags": ["inside", "desk", "sign"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "074db631b159", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.6957, "diversity_score": 0.1043, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_074db631b159", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_074db631b159/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25893801, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6993162, "qwen_elapsed_seconds": 5.114, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/record.json"} +{"prompt_id": "multi_object_relations_d3_07561e275c5b", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "synthetic photorealistic image of a tool placed inside a container on a table, sharp focus to distinguish nested objects", "target_evidence": "tool inside container", "expected_visual_elements": ["tool", "container", "table", "lid"], "negative_prompt": "blurry, low contrast, unclear boundaries", "diversity_tags": ["inside", "container", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "07561e275c5b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5909, "diversity_score": 0.2091, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_07561e275c5b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_07561e275c5b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21728305, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5979448, "qwen_elapsed_seconds": 5.139, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/record.json"} +{"prompt_id": "multi_object_relations_d3_269228cb8c34", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "synthetic photorealistic setup with a box on top of a shelf, and a lamp behind the box, clear foreground separation", "target_evidence": "box on shelf", "expected_visual_elements": ["box", "shelf", "lamp", "background"], "negative_prompt": "overlapping, low detail, weak lighting", "diversity_tags": ["on top of", "shelf", "lamp"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "269228cb8c34", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.75, "diversity_score": 0.05, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_269228cb8c34", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_269228cb8c34/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26245538, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5381559, "qwen_elapsed_seconds": 5.172, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/record.json"} +{"prompt_id": "multi_object_relations_d3_2a425cc15ef7", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "synthetic photorealistic view of a mannequin with a bag on its shoulder, and a box on top of the bag, clear visual separation", "target_evidence": "bag on shoulder", "expected_visual_elements": ["mannequin", "bag", "box", "shoulder"], "negative_prompt": "overlapping, low resolution, weak lighting", "diversity_tags": ["on top of", "bag", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2a425cc15ef7", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5909, "diversity_score": 0.2091, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_2a425cc15ef7", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24417931, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8199010, "qwen_elapsed_seconds": 5.15, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/record.json"} +{"prompt_id": "multi_object_relations_d3_3e33c1f92393", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "synthetic photorealistic setup with a mannequin sitting on a stool, and a bag on top of the stool, sharp focus for clear relation", "target_evidence": "bag on stool", "expected_visual_elements": ["mannequin", "stool", "bag", "seat"], "negative_prompt": "blurry, low contrast, unclear positioning", "diversity_tags": ["on top of", "stool", "bag"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "3e33c1f92393", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_3e33c1f92393", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_3e33c1f92393/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20705994, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1618873, "qwen_elapsed_seconds": 5.169, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/record.json"} +{"prompt_id": "multi_object_relations_d3_427f77eb50cf", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "synthetic photorealistic setup with a bag inside a container on a table, sharp focus to distinguish the nested objects clearly", "target_evidence": "bag inside container", "expected_visual_elements": ["bag", "container", "table", "container lid"], "negative_prompt": "overlapping, low resolution, unclear boundaries", "diversity_tags": ["inside", "container", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "427f77eb50cf", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.5909, "diversity_score": 0.2091, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_427f77eb50cf", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_427f77eb50cf/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24940400, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9012894, "qwen_elapsed_seconds": 5.113, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/record.json"} +{"prompt_id": "multi_object_relations_d3_4356499ba7ce", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "Synthetic photorealistic view of a mannequin holding a sign with a tool, the sign and tool clearly visible with clean boundaries.", "target_evidence": "mannequin holding sign with tool", "expected_visual_elements": ["mannequin", "sign", "tool"], "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", "diversity_tags": ["holding", "sign", "tool"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4356499ba7ce", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5455, "diversity_score": 0.2545, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_4356499ba7ce", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_4356499ba7ce/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27164474, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2502063, "qwen_elapsed_seconds": 5.07, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/record.json"} +{"prompt_id": "multi_object_relations_d3_4538c668915e", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "Synthetic photorealistic setup with a tool inside a box under a shelf, the tool clearly visible with high contrast.", "target_evidence": "tool inside box under shelf", "expected_visual_elements": ["tool", "box", "shelf"], "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", "diversity_tags": ["inside", "under", "multi-step"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4538c668915e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_4538c668915e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_4538c668915e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29243261, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 986467, "qwen_elapsed_seconds": 5.17, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/record.json"} +{"prompt_id": "multi_object_relations_d3_5555f1e89a14", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "Synthetic photorealistic setup with a box on top of a shelf, the box clearly visible and separated from the shelf edges.", "target_evidence": "box on shelf", "expected_visual_elements": ["box", "shelf"], "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", "diversity_tags": ["on top of", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5555f1e89a14", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.6, "diversity_score": 0.2, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_5555f1e89a14", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_5555f1e89a14/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20727461, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6278145, "qwen_elapsed_seconds": 5.214, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/record.json"} +{"prompt_id": "multi_object_relations_d3_571cf3225cf2", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "Synthetic photorealistic scene with a book under a table, the book clearly visible and separated from the table surface.", "target_evidence": "book under table", "expected_visual_elements": ["book", "table"], "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", "diversity_tags": ["under", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "571cf3225cf2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.4545, "diversity_score": 0.3455, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_571cf3225cf2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_571cf3225cf2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24030301, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9123441, "qwen_elapsed_seconds": 5.195, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/record.json"} +{"prompt_id": "multi_object_relations_d3_6a8fa33182a5", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "synthetic photorealistic view of a mannequin holding a key, with a box placed on top of a wooden shelf, clean boundaries for clear relation identification", "target_evidence": "mannequin holding key", "expected_visual_elements": ["mannequin", "key", "wooden shelf", "box"], "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", "diversity_tags": ["holding", "shelf", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6a8fa33182a5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_6a8fa33182a5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22513879, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 572812, "qwen_elapsed_seconds": 5.207, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/record.json"} +{"prompt_id": "multi_object_relations_d3_810283167dc6", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "synthetic photorealistic scene with a mannequin holding a sign, and a tool attached to the sign, clean boundaries for clear relation", "target_evidence": "mannequin holding sign", "expected_visual_elements": ["mannequin", "sign", "tool", "attachment point"], "negative_prompt": "blurry, low contrast, poor lighting", "diversity_tags": ["holding", "sign", "tool"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "810283167dc6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.65, "diversity_score": 0.15, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_810283167dc6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_810283167dc6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23566118, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7692250, "qwen_elapsed_seconds": 5.147, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/record.json"} +{"prompt_id": "multi_object_relations_d3_8ccffd73bf0e", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "Synthetic photorealistic setup with a tool on top of a box, the tool clearly visible with high contrast.", "target_evidence": "tool on box", "expected_visual_elements": ["tool", "box"], "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", "diversity_tags": ["on top of", "box"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "8ccffd73bf0e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_8ccffd73bf0e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28326691, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3831477, "qwen_elapsed_seconds": 5.107, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/record.json"} +{"prompt_id": "multi_object_relations_d3_cc3de276f8b9", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "Synthetic photorealistic view of a mannequin standing on a box, with a container on top of the box, clean foreground and sharp details.", "target_evidence": "mannequin standing on box", "expected_visual_elements": ["mannequin", "box", "container"], "negative_prompt": "overlapping objects, unrealistic colors, low detail", "diversity_tags": ["standing on", "box", "container"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cc3de276f8b9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_cc3de276f8b9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23327583, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5783738, "qwen_elapsed_seconds": 5.164, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/record.json"} +{"prompt_id": "multi_object_relations_d3_cdfc1eaab7f0", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "synthetic photorealistic image of a tool attached to a sign, with a clear line of sight between the objects for relation clarity", "target_evidence": "tool attached to sign", "expected_visual_elements": ["tool", "sign", "metal pole", "attachment point"], "negative_prompt": "ambiguous positioning, poor lighting, blurry edges", "diversity_tags": ["attached-to", "sign", "tool"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cdfc1eaab7f0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5217, "diversity_score": 0.2783, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_cdfc1eaab7f0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28200127, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2247326, "qwen_elapsed_seconds": 5.11, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/record.json"} +{"prompt_id": "multi_object_relations_d3_db89b92548af", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "Synthetic photorealistic scene with a mannequin holding a tool, placed on a shelf with a box behind it, clean boundaries and sharp focus.", "target_evidence": "mannequin holding tool", "expected_visual_elements": ["mannequin", "tool", "shelf", "box"], "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", "diversity_tags": ["holding", "shelf", "box"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "db89b92548af", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.6522, "diversity_score": 0.1478, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_db89b92548af", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_db89b92548af/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28352076, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2392804, "qwen_elapsed_seconds": 5.219, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/record.json"} +{"prompt_id": "multi_object_relations_d3_dc946d1ff598", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "Synthetic photorealistic image of a mannequin with a bag on its shoulder, the bag clearly visible with clean separation.", "target_evidence": "mannequin with bag on shoulder", "expected_visual_elements": ["mannequin", "bag"], "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", "diversity_tags": ["on top of", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "dc946d1ff598", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5455, "diversity_score": 0.2545, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_dc946d1ff598", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_dc946d1ff598/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23215044, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1209501, "qwen_elapsed_seconds": 5.144, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/record.json"} +{"prompt_id": "multi_object_relations_d3_e0705ad77721", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "Synthetic photorealistic scene with a mannequin holding a key, the key clearly visible in front of the mannequin's hand.", "target_evidence": "mannequin holding key", "expected_visual_elements": ["mannequin", "key"], "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", "diversity_tags": ["holding", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e0705ad77721", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.5217, "diversity_score": 0.2783, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_e0705ad77721", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_e0705ad77721/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28312520, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 567252, "qwen_elapsed_seconds": 5.227, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/record.json"} +{"prompt_id": "multi_object_relations_d3_e2e5bfbc9375", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "Synthetic photorealistic image with a tool attached to a tool rack, placed under a sign, clear foreground and high contrast.", "target_evidence": "tool attached to rack", "expected_visual_elements": ["tool", "tool rack", "sign"], "negative_prompt": "blurred edges, low lighting, unrealistic textures", "diversity_tags": ["attached to", "tool rack", "sign"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e2e5bfbc9375", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image with a", "max_similarity_to_other_prompt": 0.5455, "diversity_score": 0.2545, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_e2e5bfbc9375", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29470029, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2084011, "qwen_elapsed_seconds": 5.228, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/record.json"} +{"prompt_id": "multi_object_relations_d3_f711a1758ab2", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 3, "image_prompt": "Synthetic photorealistic scene with a mannequin holding a box, which contains a key, clear boundaries and sharp focus.", "target_evidence": "mannequin holding box containing key", "expected_visual_elements": ["mannequin", "box", "key"], "negative_prompt": "overlapping elements, low resolution, unclear objects", "diversity_tags": ["holding", "box", "key"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f711a1758ab2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.6, "diversity_score": 0.2, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d3_f711a1758ab2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_f711a1758ab2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20526890, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 979867, "qwen_elapsed_seconds": 5.177, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/record.json"} +{"prompt_id": "multi_object_relations_d4_051ad7519d81", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic setup with a bag on a desk, a pen inside the bag, and a notebook on top of the desk, high contrast, clear foreground separation", "target_evidence": "pen inside bag", "expected_visual_elements": ["bag", "pen", "notebook", "desk"], "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", "diversity_tags": ["inside", "bag", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "051ad7519d81", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.7391, "diversity_score": 0.0609, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_051ad7519d81", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_051ad7519d81/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20416768, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7510431, "qwen_elapsed_seconds": 5.209, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/record.json"} +{"prompt_id": "multi_object_relations_d4_0d4dcd9caae2", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic image with a tool attached to a sign, sharp focus and clean edges for visual legibility of the relation", "target_evidence": "tool attached to sign", "expected_visual_elements": ["tool", "sign", "wall"], "negative_prompt": "low resolution, background distractions, soft focus", "diversity_tags": ["attached-to", "sign", "tool"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0d4dcd9caae2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image with a", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_0d4dcd9caae2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28008680, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9649498, "qwen_elapsed_seconds": 5.184, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/record.json"} +{"prompt_id": "multi_object_relations_d4_11d2a4093355", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic image of a box with a book inside, a mug on the table, and a sign next to the box, high contrast, clear foreground separation", "target_evidence": "book inside box", "expected_visual_elements": ["box", "book", "mug", "sign", "table"], "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", "diversity_tags": ["inside", "box", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "11d2a4093355", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_11d2a4093355", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_11d2a4093355/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23095593, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6489438, "qwen_elapsed_seconds": 5.117, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/record.json"} +{"prompt_id": "multi_object_relations_d4_1b6fe1b04f06", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic setup with a tool placed on top of a box on a shelf, clear foreground separation and high contrast", "target_evidence": "tool on top of box", "expected_visual_elements": ["tool", "box", "shelf"], "negative_prompt": "distorted perspective, low contrast, unclear edges", "diversity_tags": ["on top of", "box", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1b6fe1b04f06", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.8, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_1b6fe1b04f06", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29886781, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5979193, "qwen_elapsed_seconds": 5.188, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/record.json"} +{"prompt_id": "multi_object_relations_d4_244ce55796c6", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic setup with a box on top of a shelf, high contrast, clean boundaries, and distinct foreground objects.", "target_evidence": "box on top of shelf", "expected_visual_elements": ["box", "shelf"], "negative_prompt": "blurry, low contrast, overlapping objects, real people", "diversity_tags": ["on top of", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "244ce55796c6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.6364, "diversity_score": 0.1636, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_244ce55796c6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_244ce55796c6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29678577, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2689398, "qwen_elapsed_seconds": 5.178, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/record.json"} +{"prompt_id": "multi_object_relations_d4_3bf4f5d535e5", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic image of a mannequin with a bag on its shoulder, a pen inside the bag, and a sign near the mannequin, clean boundaries, sharp focus", "target_evidence": "pen inside bag", "expected_visual_elements": ["mannequin", "bag", "pen", "sign"], "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", "diversity_tags": ["inside", "bag", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "3bf4f5d535e5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.6364, "diversity_score": 0.1636, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_3bf4f5d535e5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29156616, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7859375, "qwen_elapsed_seconds": 5.232, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/record.json"} +{"prompt_id": "multi_object_relations_d4_3d1b4c30ae72", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic scene with a mannequin holding a toolbox, a wrench clearly in its grip, and a screwdriver nearby on a shelf, clean boundaries, sharp focus", "target_evidence": "mannequin holding toolbox", "expected_visual_elements": ["mannequin", "toolbox", "wrench", "shelf"], "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", "diversity_tags": ["holding", "mannequin", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "3d1b4c30ae72", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.56, "diversity_score": 0.24, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_3d1b4c30ae72", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22082823, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6994313, "qwen_elapsed_seconds": 5.143, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/record.json"} +{"prompt_id": "multi_object_relations_d4_541f29108785", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic setup with a mannequin standing behind a chair, clean edges and sharp focus to emphasize the behind relation", "target_evidence": "mannequin behind chair", "expected_visual_elements": ["mannequin", "chair", "floor"], "negative_prompt": "overlapping elements, poor lighting, low contrast", "diversity_tags": ["behind", "chair", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "541f29108785", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_541f29108785", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_541f29108785/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24655905, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2320017, "qwen_elapsed_seconds": 5.142, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/record.json"} +{"prompt_id": "multi_object_relations_d4_57a5e1f493b5", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic scene with a mannequin holding a bag, a key attached to the bag, and a tool on the desk, clean boundaries, sharp focus", "target_evidence": "mannequin holding bag", "expected_visual_elements": ["mannequin", "bag", "key", "tool", "desk"], "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", "diversity_tags": ["holding", "mannequin", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "57a5e1f493b5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.7143, "diversity_score": 0.0857, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_57a5e1f493b5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27538672, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3645504, "qwen_elapsed_seconds": 5.219, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/record.json"} +{"prompt_id": "multi_object_relations_d4_58ff6e8b16b1", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a mannequin with a tool attached to its arm, clean boundaries, high contrast, and clear foreground.", "target_evidence": "tool attached to mannequin", "expected_visual_elements": ["mannequin", "tool"], "negative_prompt": "blurry, low contrast, overlapping objects, real people", "diversity_tags": ["attached to", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "58ff6e8b16b1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.75, "diversity_score": 0.05, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_58ff6e8b16b1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26440915, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9559365, "qwen_elapsed_seconds": 5.213, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/record.json"} +{"prompt_id": "multi_object_relations_d4_5b222069f69d", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a box on a desk with a tool nearby, distinct foreground separation, and high contrast.", "target_evidence": "box on desk", "expected_visual_elements": ["box", "desk"], "negative_prompt": "blurry, low contrast, overlapping objects, real people", "diversity_tags": ["on top of", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5b222069f69d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.619, "diversity_score": 0.181, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_5b222069f69d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_5b222069f69d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22137842, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9803052, "qwen_elapsed_seconds": 5.142, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/record.json"} +{"prompt_id": "multi_object_relations_d4_62ec93cecea4", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic setup with a tool on a shelf, a wrench on top of the tool, and a bag near the shelf, clean boundaries, sharp focus", "target_evidence": "wrench on top of tool", "expected_visual_elements": ["tool", "wrench", "bag", "shelf"], "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", "diversity_tags": ["on top of", "tool", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "62ec93cecea4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.619, "diversity_score": 0.181, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_62ec93cecea4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_62ec93cecea4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20476669, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6514423, "qwen_elapsed_seconds": 5.125, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/record.json"} +{"prompt_id": "multi_object_relations_d4_64ae9cc06dca", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic setup with a bag placed inside a container on a desk, high contrast and clear foreground separation for relation analysis", "target_evidence": "bag inside container", "expected_visual_elements": ["bag", "container", "desk"], "negative_prompt": "distorted perspective, unclear boundaries, reflections", "diversity_tags": ["inside", "container", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "64ae9cc06dca", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_64ae9cc06dca", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26751141, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5779002, "qwen_elapsed_seconds": 5.15, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/record.json"} +{"prompt_id": "multi_object_relations_d4_6ad705a61b48", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic scene with a book on top of a box on a table, clear foreground separation and high contrast for relation inspection", "target_evidence": "book on top of box", "expected_visual_elements": ["book", "box", "table"], "negative_prompt": "faded colors, background clutter, unclear lighting", "diversity_tags": ["on top of", "box", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6ad705a61b48", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.175, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_6ad705a61b48", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_6ad705a61b48/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26682356, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5793693, "qwen_elapsed_seconds": 5.175, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/record.json"} +{"prompt_id": "multi_object_relations_d4_6bdcb20a1fd4", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic scene with a mannequin standing behind a chair, a lamp on the desk, and a cup on the chair, clear foreground separation, high contrast", "target_evidence": "mannequin behind chair", "expected_visual_elements": ["mannequin", "chair", "lamp", "cup", "desk"], "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", "diversity_tags": ["behind", "mannequin", "chair"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6bdcb20a1fd4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.6087, "diversity_score": 0.1913, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_6bdcb20a1fd4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23139165, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2585804, "qwen_elapsed_seconds": 5.112, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/record.json"} +{"prompt_id": "multi_object_relations_d4_8ceda8b84aa3", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic setup with a container on a shelf, a bottle inside the container, and a tool on the shelf, high contrast, clear foreground separation", "target_evidence": "bottle inside container", "expected_visual_elements": ["container", "bottle", "tool", "shelf"], "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", "diversity_tags": ["inside", "container", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "8ceda8b84aa3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.6364, "diversity_score": 0.1636, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_8ceda8b84aa3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28209408, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9351229, "qwen_elapsed_seconds": 5.178, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/record.json"} +{"prompt_id": "multi_object_relations_d4_a48ec906a073", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic scene with a bag on a mannequin's shoulder, clean separation and sharp focus for the held-by relation", "target_evidence": "bag on mannequin's shoulder", "expected_visual_elements": ["bag", "mannequin", "shoulder"], "negative_prompt": "blurred edges, low resolution, cluttered background", "diversity_tags": ["held-by", "mannequin", "bag"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a48ec906a073", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.6, "diversity_score": 0.2, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_a48ec906a073", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_a48ec906a073/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29554811, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5188663, "qwen_elapsed_seconds": 5.098, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/record.json"} +{"prompt_id": "multi_object_relations_d4_dc8bf3cb3eab", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic scene with a mannequin holding a box on a shelf, clean boundaries and sharp focus for clear relation inspection", "target_evidence": "mannequin holding box", "expected_visual_elements": ["mannequin", "box", "shelf"], "negative_prompt": "blurry details, overlapping objects, low contrast", "diversity_tags": ["holding", "shelf", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "dc8bf3cb3eab", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_dc8bf3cb3eab", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22020640, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7567140, "qwen_elapsed_seconds": 5.225, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/record.json"} +{"prompt_id": "multi_object_relations_d4_e00ad2c64904", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic scene with a container inside a bag on a desk, clear boundaries and high contrast for relation analysis", "target_evidence": "container inside bag", "expected_visual_elements": ["container", "bag", "desk"], "negative_prompt": "distorted perspective, low contrast, unclear edges", "diversity_tags": ["inside", "bag", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e00ad2c64904", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_e00ad2c64904", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_e00ad2c64904/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28208223, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9235420, "qwen_elapsed_seconds": 5.099, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/record.json"} +{"prompt_id": "multi_object_relations_d4_e2ae8a1373cb", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 4, "image_prompt": "synthetic photorealistic image of a box with a key attached to its side, a small plant on the table, and a sign next to the box, clean boundaries, sharp focus", "target_evidence": "key attached to box", "expected_visual_elements": ["box", "key", "plant", "sign", "table"], "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", "diversity_tags": ["attached-to", "box", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e2ae8a1373cb", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d4_e2ae8a1373cb", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26753698, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 161678, "qwen_elapsed_seconds": 5.158, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/record.json"} +{"prompt_id": "multi_object_relations_d5_11fc8e7a0239", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic image of a bag under a chair, with a bookshelf in the background, clear foreground separation", "target_evidence": "bag under chair", "expected_visual_elements": ["bag", "chair", "bookshelf", "clear foreground"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["under", "bag", "furniture"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "11fc8e7a0239", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.4091, "diversity_score": 0.3909, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_11fc8e7a0239", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25246698, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2451426, "qwen_elapsed_seconds": 5.167, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/record.json"} +{"prompt_id": "multi_object_relations_d5_34c2ccf3b5ef", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "Synthetic photorealistic setup with a bag placed on top of a table, clean boundaries and high contrast for visual clarity.", "target_evidence": "bag on table", "expected_visual_elements": ["bag", "table", "clear boundaries"], "negative_prompt": "blurry, low contrast, overlapping objects, real people, faces", "diversity_tags": ["on top of", "table", "bag"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "34c2ccf3b5ef", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.5909, "diversity_score": 0.2091, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_34c2ccf3b5ef", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22483845, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2615726, "qwen_elapsed_seconds": 5.109, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/record.json"} +{"prompt_id": "multi_object_relations_d5_4f451cd1db1d", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic setup with a box on top of a shelf, a sign next to it, sharp focus on the relation between them", "target_evidence": "box on top of shelf", "expected_visual_elements": ["box", "shelf", "sign", "sharp focus"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["on top of", "box", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4f451cd1db1d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_4f451cd1db1d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26509252, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2735174, "qwen_elapsed_seconds": 5.158, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/record.json"} +{"prompt_id": "multi_object_relations_d5_5db805d7dab7", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic setup with a tool on top of a mannequin's head, a box beside it, high contrast for visual clarity", "target_evidence": "tool on top of mannequin", "expected_visual_elements": ["tool", "mannequin", "box", "high contrast"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["on top of", "tool", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5db805d7dab7", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.5217, "diversity_score": 0.2783, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_5db805d7dab7", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_5db805d7dab7/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20942630, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5378640, "qwen_elapsed_seconds": 5.126, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/record.json"} +{"prompt_id": "multi_object_relations_d5_5eae587b509d", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "Synthetic photorealistic view with a mannequin with a bag attached to its arm, clean boundaries and high contrast.", "target_evidence": "bag attached to mannequin", "expected_visual_elements": ["mannequin", "bag", "clean boundaries"], "negative_prompt": "blurry, low contrast, overlapping objects, real people, faces", "diversity_tags": ["attached to", "mannequin", "bag"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5eae587b509d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view with a", "max_similarity_to_other_prompt": 0.75, "diversity_score": 0.25, "repeated_opening": false, "review": {"prompt_id": "multi_object_relations_d5_5eae587b509d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_5eae587b509d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21052723, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2471002, "qwen_elapsed_seconds": 5.111, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/record.json"} +{"prompt_id": "multi_object_relations_d5_5eeca9085072", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "Synthetic photorealistic scene with a mannequin holding a box on a shelf, clear foreground separation and sharp focus.", "target_evidence": "mannequin holding box", "expected_visual_elements": ["mannequin", "box", "shelf"], "negative_prompt": "blurry, low contrast, overlapping objects, real people, faces", "diversity_tags": ["holding", "shelf", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5eeca9085072", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.7368, "diversity_score": 0.0632, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_5eeca9085072", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_5eeca9085072/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27573616, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 944517, "qwen_elapsed_seconds": 5.115, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/record.json"} +{"prompt_id": "multi_object_relations_d5_63e8915777db", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic scene with a mannequin standing next to a box with clear visual legibility and sharp focus.", "target_evidence": "mannequin next to box", "expected_visual_elements": ["mannequin", "box", "sharp focus"], "negative_prompt": "low contrast, ambiguous placement, soft edges", "diversity_tags": ["next to", "box", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "63e8915777db", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.5238, "diversity_score": 0.2762, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_63e8915777db", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_63e8915777db/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28877290, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5370677, "qwen_elapsed_seconds": 5.181, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/record.json"} +{"prompt_id": "multi_object_relations_d5_6e43b96c2267", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic scene with a container behind a desk, a tool on the desk, clear foreground separation for relation", "target_evidence": "container behind desk", "expected_visual_elements": ["container", "desk", "tool", "clear separation"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["behind", "container", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6e43b96c2267", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_6e43b96c2267", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_6e43b96c2267/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29377294, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 735298, "qwen_elapsed_seconds": 5.157, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/record.json"} +{"prompt_id": "multi_object_relations_d5_97c8654f9f45", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "Synthetic photorealistic view of a mannequin holding a toolbox, with a wrench on top of the toolbox and a screwdriver inside the toolbox, clean boundaries and sharp focus.", "target_evidence": "mannequin holding toolbox with wrench on top and screwdriver inside", "expected_visual_elements": ["mannequin", "toolbox", "wrench", "screwdriver"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks, real people, faces", "diversity_tags": ["holding", "on top of", "inside", "mannequin"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "97c8654f9f45", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.6087, "diversity_score": 0.1913, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_97c8654f9f45", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_97c8654f9f45/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20498959, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4556878, "qwen_elapsed_seconds": 5.194, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/record.json"} +{"prompt_id": "multi_object_relations_d5_98689667028e", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a tool resting on top of a box with clear visual separation and high contrast.", "target_evidence": "tool on top of box", "expected_visual_elements": ["tool", "box", "high contrast"], "negative_prompt": "low detail, overlapping elements, soft focus", "diversity_tags": ["on top of", "box", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "98689667028e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_98689667028e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_98689667028e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25981097, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6143210, "qwen_elapsed_seconds": 5.118, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/record.json"} +{"prompt_id": "multi_object_relations_d5_99ab20067c80", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "Synthetic photorealistic scene with a sign attached to a wall, high contrast and clean edges for visual separation.", "target_evidence": "sign attached to wall", "expected_visual_elements": ["sign", "wall", "clean edges"], "negative_prompt": "blurry, low contrast, overlapping objects, real people, faces", "diversity_tags": ["attached to", "wall", "sign"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "99ab20067c80", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.75, "diversity_score": 0.05, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_99ab20067c80", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_99ab20067c80/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23192110, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6069800, "qwen_elapsed_seconds": 5.205, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/record.json"} +{"prompt_id": "multi_object_relations_d5_a2bed91d138b", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a mannequin with a sign attached to its arm, a tool on the ground nearby, high contrast for clarity", "target_evidence": "sign attached to mannequin", "expected_visual_elements": ["mannequin", "sign", "tool", "high contrast"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["attached to", "mannequin", "tool"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a2bed91d138b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.64, "diversity_score": 0.16, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_a2bed91d138b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_a2bed91d138b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29882024, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 298442, "qwen_elapsed_seconds": 5.139, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/record.json"} +{"prompt_id": "multi_object_relations_d5_aac42d59f553", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "Synthetic photorealistic scene with a bag on a shelf, a key inside the bag, and a notebook on top of the bag, high contrast and clear foreground separation.", "target_evidence": "bag on shelf with key inside and notebook on top", "expected_visual_elements": ["bag", "shelf", "key", "notebook"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks, real people, faces", "diversity_tags": ["on top of", "inside", "shelf", "bag"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "aac42d59f553", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.7391, "diversity_score": 0.0609, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_aac42d59f553", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_aac42d59f553/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27540329, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1127660, "qwen_elapsed_seconds": 5.207, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/record.json"} +{"prompt_id": "multi_object_relations_d5_afeb8e5dfdd6", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic setup with a bag placed inside a container on a desk, ensuring sharp focus on the relation.", "target_evidence": "bag inside container", "expected_visual_elements": ["bag", "container", "desk"], "negative_prompt": "distorted perspective, low resolution, ambiguous placement", "diversity_tags": ["inside", "container", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "afeb8e5dfdd6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.5455, "diversity_score": 0.2545, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_afeb8e5dfdd6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_afeb8e5dfdd6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26874572, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_afeb8e5dfdd6/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_afeb8e5dfdd6/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_afeb8e5dfdd6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7817290, "qwen_elapsed_seconds": 5.197, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_afeb8e5dfdd6/record.json"} +{"prompt_id": "multi_object_relations_d5_b22378e741c5", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "Synthetic photorealistic setup with a container on a desk, a pen attached to the container, and a ruler behind the container, sharp focus and clean boundaries.", "target_evidence": "container on desk with pen attached and ruler behind", "expected_visual_elements": ["container", "desk", "pen", "ruler"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks, real people, faces", "diversity_tags": ["attached to", "behind", "desk", "container"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b22378e741c5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.56, "diversity_score": 0.24, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_b22378e741c5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_b22378e741c5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25523229, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_b22378e741c5/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_b22378e741c5/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_b22378e741c5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6593462, "qwen_elapsed_seconds": 5.2, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_b22378e741c5/record.json"} +{"prompt_id": "multi_object_relations_d5_b4810c0388f4", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic scene with a bag on top of a table, a chair next to it, sharp focus on the relation between them", "target_evidence": "bag on top of table", "expected_visual_elements": ["bag", "table", "chair", "sharp focus"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["on top of", "table", "furniture"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b4810c0388f4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_b4810c0388f4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_b4810c0388f4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23514411, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_b4810c0388f4/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_b4810c0388f4/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_b4810c0388f4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6181756, "qwen_elapsed_seconds": 5.129, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_b4810c0388f4/record.json"} +{"prompt_id": "multi_object_relations_d5_caac2354dcf6", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic image of a mannequin with a tool attached to its back, a bag on the floor, high contrast for clarity", "target_evidence": "tool attached to mannequin", "expected_visual_elements": ["mannequin", "tool", "bag", "high contrast"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["attached to", "tool", "ground object"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "caac2354dcf6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.64, "diversity_score": 0.16, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_caac2354dcf6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_caac2354dcf6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21340450, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_caac2354dcf6/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_caac2354dcf6/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_caac2354dcf6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2421276, "qwen_elapsed_seconds": 5.137, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_caac2354dcf6/record.json"} +{"prompt_id": "multi_object_relations_d5_d56b4376a3c5", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic setup with a container inside a box, both on a shelf, clean boundaries for clear separation", "target_evidence": "container inside box", "expected_visual_elements": ["container", "box", "shelf", "clean boundaries"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["inside", "container", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d56b4376a3c5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup with a", "max_similarity_to_other_prompt": 0.4783, "diversity_score": 0.3217, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_d56b4376a3c5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_d56b4376a3c5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 30088649, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_d56b4376a3c5/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_d56b4376a3c5/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_d56b4376a3c5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3792902, "qwen_elapsed_seconds": 5.144, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_d56b4376a3c5/record.json"} +{"prompt_id": "multi_object_relations_d5_f2d1e004d6ca", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a mannequin holding a toolbox, with a ladder leaning against a wall behind, clear foreground separation", "target_evidence": "mannequin holding toolbox", "expected_visual_elements": ["mannequin", "toolbox", "ladder", "wall"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["holding", "mannequin", "background object"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f2d1e004d6ca", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.375, "diversity_score": 0.425, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_f2d1e004d6ca", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_f2d1e004d6ca/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27812676, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_f2d1e004d6ca/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_f2d1e004d6ca/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_f2d1e004d6ca/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 454858, "qwen_elapsed_seconds": 5.125, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_f2d1e004d6ca/record.json"} +{"prompt_id": "multi_object_relations_d5_f6cae1eb386f", "run_name": "flux2_qil_decomposition_pilot", "category": "multi_object_relations", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a mannequin holding a bottle, with a bag on the ground nearby, sharp focus on the relation", "target_evidence": "mannequin holding bottle", "expected_visual_elements": ["mannequin", "bottle", "bag", "sharp focus"], "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", "diversity_tags": ["holding", "bottle", "ground object"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f6cae1eb386f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "multi_object_relations_d5_f6cae1eb386f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_f6cae1eb386f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21598028, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/multi_object_relations/multi_object_relations_d5_f6cae1eb386f/layers/layer_00.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_f6cae1eb386f/layers/layer_01.png", "layers-x/k3/multi_object_relations/multi_object_relations_d5_f6cae1eb386f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 804042, "qwen_elapsed_seconds": 5.146, "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_f6cae1eb386f/record.json"} +{"prompt_id": "occlusion_overlap_d1_0cd43d9133a7", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "Synthetic photorealistic view of a silver spoon overlapping a green leaf on a white countertop, clean boundaries, high contrast between objects", "target_evidence": "green leaf", "expected_visual_elements": ["silver spoon", "green leaf", "white countertop"], "negative_prompt": "fuzzy outlines, low resolution, poor depth ordering, cluttered background", "diversity_tags": ["utensil", "plant", "countertop"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0cd43d9133a7", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_0cd43d9133a7", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_0cd43d9133a7/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22750346, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_0cd43d9133a7/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_0cd43d9133a7/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_0cd43d9133a7/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1303226, "qwen_elapsed_seconds": 5.111, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_0cd43d9133a7/record.json"} +{"prompt_id": "occlusion_overlap_d1_12a77ce170eb", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "synthetic photorealistic view of a green ceramic plant pot overlapping a white ceramic mug on a white countertop, high contrast, clear depth", "target_evidence": "green ceramic plant pot", "expected_visual_elements": ["green ceramic plant pot", "white ceramic mug", "white countertop"], "negative_prompt": "blurry, overlapping edges, low resolution", "diversity_tags": ["ceramic", "plant", "transparent overlap"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "12a77ce170eb", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_12a77ce170eb", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_12a77ce170eb/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27977077, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_12a77ce170eb/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_12a77ce170eb/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_12a77ce170eb/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4758443, "qwen_elapsed_seconds": 5.085, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_12a77ce170eb/record.json"} +{"prompt_id": "occlusion_overlap_d1_193f423d5fb8", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "synthetic photorealistic setup of a brown leather wallet partially covered by a blue notebook on a wooden floor, clean boundaries, sharp focus", "target_evidence": "brown leather wallet", "expected_visual_elements": ["brown leather wallet", "blue notebook", "wooden floor"], "negative_prompt": "low detail, low contrast, cluttered background", "diversity_tags": ["leather", "paper", "partial occlusion"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "193f423d5fb8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4828, "diversity_score": 0.3172, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_193f423d5fb8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_193f423d5fb8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20461455, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_193f423d5fb8/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_193f423d5fb8/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_193f423d5fb8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 631499, "qwen_elapsed_seconds": 5.122, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_193f423d5fb8/record.json"} +{"prompt_id": "occlusion_overlap_d1_223bdbe7f217", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "synthetic photorealistic setup of a yellow metal key partially covered by a black fabric cloth on a dark wooden shelf, clean edges, sharp lighting", "target_evidence": "yellow metal key", "expected_visual_elements": ["yellow metal key", "black fabric cloth", "dark wooden shelf"], "negative_prompt": "foggy, soft focus, cluttered background", "diversity_tags": ["metal", "fabric", "partial occlusion"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "223bdbe7f217", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.6296, "diversity_score": 0.1704, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_223bdbe7f217", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_223bdbe7f217/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28326971, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_223bdbe7f217/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_223bdbe7f217/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_223bdbe7f217/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3241622, "qwen_elapsed_seconds": 5.171, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_223bdbe7f217/record.json"} +{"prompt_id": "occlusion_overlap_d1_3c6c5154e10a", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "Synthetic photorealistic view of a red pen partially covered by a blue cloth on a white wall, sharp focus, clean foreground separation", "target_evidence": "red pen", "expected_visual_elements": ["red pen", "blue cloth", "white wall"], "negative_prompt": "fuzzy edges, low resolution, overlapping shadows, cluttered background", "diversity_tags": ["tool", "fabric", "wall"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "3c6c5154e10a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_3c6c5154e10a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_3c6c5154e10a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29325269, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_3c6c5154e10a/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_3c6c5154e10a/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_3c6c5154e10a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 861995, "qwen_elapsed_seconds": 5.108, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_3c6c5154e10a/record.json"} +{"prompt_id": "occlusion_overlap_d1_528e51b67416", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "synthetic photorealistic scene of a white ceramic teacup partially hidden under a brown coffee mug on a dark wooden table, visible fraction over 60%, clean edges", "target_evidence": "white ceramic teacup", "expected_visual_elements": ["white ceramic teacup", "brown coffee mug", "dark wooden table"], "negative_prompt": "blurry, soft focus, low contrast", "diversity_tags": ["ceramic", "coffee", "partial occlusion"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "528e51b67416", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.6207, "diversity_score": 0.1793, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_528e51b67416", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_528e51b67416/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29611464, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_528e51b67416/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_528e51b67416/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_528e51b67416/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2168938, "qwen_elapsed_seconds": 5.122, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_528e51b67416/record.json"} +{"prompt_id": "occlusion_overlap_d1_6038b8af716d", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "synthetic photorealistic scene of a red ceramic mug partially hidden under a green wooden bowl on a wooden table, clean boundaries, sharp focus", "target_evidence": "red ceramic mug", "expected_visual_elements": ["red ceramic mug", "green wooden bowl", "wooden table"], "negative_prompt": "blurry, low contrast, chaotic composition", "diversity_tags": ["ceramic", "wooden", "partial occlusion"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6038b8af716d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_6038b8af716d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_6038b8af716d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22055221, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_6038b8af716d/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_6038b8af716d/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_6038b8af716d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6543095, "qwen_elapsed_seconds": 5.112, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_6038b8af716d/record.json"} +{"prompt_id": "occlusion_overlap_d1_7520f214b992", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "Synthetic photorealistic scene of a blue bottle partially obscured by a green plant on a stone shelf, clean boundaries, clear foreground separation", "target_evidence": "blue bottle", "expected_visual_elements": ["blue bottle", "green plant", "stone shelf"], "negative_prompt": "blurry textures, low contrast, overlapping details, cluttered background", "diversity_tags": ["container", "plant", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7520f214b992", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_7520f214b992", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_7520f214b992/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29930326, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_7520f214b992/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_7520f214b992/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_7520f214b992/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4391899, "qwen_elapsed_seconds": 5.216, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_7520f214b992/record.json"} +{"prompt_id": "occlusion_overlap_d1_7819585f2251", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "synthetic photorealistic scene with a silver laptop partially hidden under a black notebook on a desk, visible fraction over 70%, clean separation", "target_evidence": "silver laptop", "expected_visual_elements": ["silver laptop", "black notebook", "desk"], "negative_prompt": "low contrast, soft shadows, unclear boundaries", "diversity_tags": ["electronics", "paper", "partial occlusion"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7819585f2251", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.4828, "diversity_score": 0.3172, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_7819585f2251", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_7819585f2251/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21561520, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_7819585f2251/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_7819585f2251/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_7819585f2251/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1312823, "qwen_elapsed_seconds": 5.135, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_7819585f2251/record.json"} +{"prompt_id": "occlusion_overlap_d1_946e2f5478d7", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "Synthetic photorealistic setup of a green mug partially hidden under a black book on a wooden shelf, clear foreground separation, high contrast", "target_evidence": "green mug", "expected_visual_elements": ["green mug", "black book", "wooden shelf"], "negative_prompt": "indistinct edges, low detail, overlapping shadows, cluttered background", "diversity_tags": ["container", "book", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "946e2f5478d7", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_946e2f5478d7", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_946e2f5478d7/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20867132, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_946e2f5478d7/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_946e2f5478d7/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_946e2f5478d7/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8846343, "qwen_elapsed_seconds": 5.224, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_946e2f5478d7/record.json"} +{"prompt_id": "occlusion_overlap_d1_9584e081f017", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "Synthetic photorealistic scene of a yellow key partially obscured by a transparent glass on a metal counter, clean boundaries, sharp focus", "target_evidence": "yellow key", "expected_visual_elements": ["yellow key", "transparent glass", "metal counter"], "negative_prompt": "blurry textures, low contrast, overlapping details, messy arrangement", "diversity_tags": ["tool", "glass", "counter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "9584e081f017", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.175, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_9584e081f017", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_9584e081f017/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23861488, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_9584e081f017/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_9584e081f017/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_9584e081f017/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3540311, "qwen_elapsed_seconds": 5.133, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_9584e081f017/record.json"} +{"prompt_id": "occlusion_overlap_d1_a8bdfa503b9c", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "synthetic photorealistic view of a blue glass vial overlapping a white plastic container on a counter, high contrast, clear foreground separation", "target_evidence": "blue glass vial", "expected_visual_elements": ["blue glass vial", "white plastic container", "counter"], "negative_prompt": "low detail, overlapping edges, low resolution", "diversity_tags": ["glass", "plastic", "transparent overlap"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a8bdfa503b9c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_a8bdfa503b9c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_a8bdfa503b9c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21201309, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_a8bdfa503b9c/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_a8bdfa503b9c/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_a8bdfa503b9c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1512340, "qwen_elapsed_seconds": 5.144, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_a8bdfa503b9c/record.json"} +{"prompt_id": "occlusion_overlap_d1_a92c60742643", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "Synthetic photorealistic view of a white cup partially covered by a brown paper towel on a tile floor, sharp focus, clean foreground separation", "target_evidence": "white cup", "expected_visual_elements": ["white cup", "brown paper towel", "tile floor"], "negative_prompt": "fuzzy edges, low resolution, overlapping shadows, chaotic composition", "diversity_tags": ["container", "material", "floor"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a92c60742643", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_a92c60742643", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_a92c60742643/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20949766, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_a92c60742643/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_a92c60742643/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_a92c60742643/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6025483, "qwen_elapsed_seconds": 5.175, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_a92c60742643/record.json"} +{"prompt_id": "occlusion_overlap_d1_ac5a38c8d409", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "Synthetic photorealistic setup of a black phone partially hidden under a white notebook on a wooden table, clear foreground separation, high contrast", "target_evidence": "black phone", "expected_visual_elements": ["black phone", "white notebook", "wooden table"], "negative_prompt": "indistinct edges, low detail, overlapping shadows, chaotic composition", "diversity_tags": ["electronics", "book", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ac5a38c8d409", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_ac5a38c8d409", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_ac5a38c8d409/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25039534, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_ac5a38c8d409/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_ac5a38c8d409/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_ac5a38c8d409/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 234922, "qwen_elapsed_seconds": 5.126, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_ac5a38c8d409/record.json"} +{"prompt_id": "occlusion_overlap_d1_d2e4c806d4bc", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "synthetic photorealistic scene of a white ceramic mug partially hidden under a black metal bowl on a wooden counter, visible fraction over 70%, clean edges", "target_evidence": "white ceramic mug", "expected_visual_elements": ["white ceramic mug", "black metal bowl", "wooden counter"], "negative_prompt": "low contrast, soft shadows, unclear boundaries", "diversity_tags": ["ceramic", "metal", "partial occlusion"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d2e4c806d4bc", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.6207, "diversity_score": 0.1793, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_d2e4c806d4bc", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_d2e4c806d4bc/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27030963, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_d2e4c806d4bc/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_d2e4c806d4bc/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_d2e4c806d4bc/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9443759, "qwen_elapsed_seconds": 5.098, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_d2e4c806d4bc/record.json"} +{"prompt_id": "occlusion_overlap_d1_d462708b3b8f", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "synthetic photorealistic setup of a red plastic bottle partially covered by a green fabric cloth on a dark wooden shelf, clean boundaries, sharp lighting", "target_evidence": "red plastic bottle", "expected_visual_elements": ["red plastic bottle", "green fabric cloth", "dark wooden shelf"], "negative_prompt": "blurry, soft focus, cluttered background", "diversity_tags": ["plastic", "fabric", "partial occlusion"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d462708b3b8f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.6296, "diversity_score": 0.1704, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_d462708b3b8f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_d462708b3b8f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21201112, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_d462708b3b8f/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_d462708b3b8f/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_d462708b3b8f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6594950, "qwen_elapsed_seconds": 5.179, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_d462708b3b8f/record.json"} +{"prompt_id": "occlusion_overlap_d1_d6d5fc960689", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "synthetic photorealistic view of a black metal spoon overlapping a white ceramic plate on a white countertop, high contrast, clear foreground", "target_evidence": "black metal spoon", "expected_visual_elements": ["black metal spoon", "white ceramic plate", "white countertop"], "negative_prompt": "low detail, overlapping edges, low resolution", "diversity_tags": ["metal", "ceramic", "transparent overlap"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d6d5fc960689", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_d6d5fc960689", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_d6d5fc960689/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28940612, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_d6d5fc960689/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_d6d5fc960689/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_d6d5fc960689/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1069842, "qwen_elapsed_seconds": 5.178, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_d6d5fc960689/record.json"} +{"prompt_id": "occlusion_overlap_d1_e7a0b1712fa5", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "Synthetic photorealistic setup of a yellow pencil partially covered by a black notebook on a gray desk, clear foreground separation, sharp focus on pencil tip", "target_evidence": "yellow pencil", "expected_visual_elements": ["yellow pencil", "black notebook", "gray desk"], "negative_prompt": "indistinct edges, low detail, overlapping shadows, messy arrangement", "diversity_tags": ["stationery", "book", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e7a0b1712fa5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4138, "diversity_score": 0.3862, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_e7a0b1712fa5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_e7a0b1712fa5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20528991, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_e7a0b1712fa5/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_e7a0b1712fa5/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_e7a0b1712fa5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6363680, "qwen_elapsed_seconds": 5.17, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_e7a0b1712fa5/record.json"} +{"prompt_id": "occlusion_overlap_d1_ea6e45868f9b", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "Synthetic photorealistic scene of a green mug partially obscured by a transparent plastic sheet on a stone counter, clean boundaries, sharp focus", "target_evidence": "green mug", "expected_visual_elements": ["green mug", "transparent plastic sheet", "stone counter"], "negative_prompt": "blurry textures, low contrast, overlapping details, messy arrangement", "diversity_tags": ["container", "material", "counter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ea6e45868f9b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.175, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_ea6e45868f9b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_ea6e45868f9b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26146721, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_ea6e45868f9b/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_ea6e45868f9b/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_ea6e45868f9b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1830457, "qwen_elapsed_seconds": 5.122, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_ea6e45868f9b/record.json"} +{"prompt_id": "occlusion_overlap_d1_f0469dddfa3e", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 1, "image_prompt": "Synthetic photorealistic scene of a red apple partially hidden under a blue ceramic mug on a wooden table, clear foreground separation, sharp focus on apple edges", "target_evidence": "red apple", "expected_visual_elements": ["red apple", "blue ceramic mug", "wooden table"], "negative_prompt": "blurry edges, low contrast, overlapping textures, chaotic composition", "diversity_tags": ["fruit", "container", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f0469dddfa3e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.7037, "diversity_score": 0.0963, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d1_f0469dddfa3e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d1_f0469dddfa3e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28016125, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d1_f0469dddfa3e/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_f0469dddfa3e/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_f0469dddfa3e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6699752, "qwen_elapsed_seconds": 5.118, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d1_f0469dddfa3e/record.json"} +{"prompt_id": "occlusion_overlap_d2_040006c74af1", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic view of a red plastic pen partially hidden under a black fabric notebook on a wooden floor, with clear foreground and sharp focus", "target_evidence": "red plastic pen", "expected_visual_elements": ["red plastic pen", "black fabric notebook", "wooden floor"], "negative_prompt": "low contrast, messy composition, overlapping textures, poor depth", "diversity_tags": ["plastic", "notebook", "floor"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "040006c74af1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.6207, "diversity_score": 0.1793, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_040006c74af1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_040006c74af1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27343690, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_040006c74af1/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_040006c74af1/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_040006c74af1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4299987, "qwen_elapsed_seconds": 5.194, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_040006c74af1/record.json"} +{"prompt_id": "occlusion_overlap_d2_10aab92cfd96", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic setup of a black ceramic vase partially covered by a white linen cloth on a stone shelf, with clear boundaries and high contrast", "target_evidence": "black ceramic vase", "expected_visual_elements": ["black ceramic vase", "white linen cloth", "stone shelf"], "negative_prompt": "low definition, poor lighting, merged objects, low depth", "diversity_tags": ["ceramic", "linen", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "10aab92cfd96", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_10aab92cfd96", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_10aab92cfd96/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 30189503, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_10aab92cfd96/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_10aab92cfd96/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_10aab92cfd96/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7735078, "qwen_elapsed_seconds": 5.13, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_10aab92cfd96/record.json"} +{"prompt_id": "occlusion_overlap_d2_1bf923ec2a92", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic view of a green metal ruler partially hidden under a blue fabric notebook on a wooden desk, with sharp focus and clear foreground separation", "target_evidence": "green metal ruler", "expected_visual_elements": ["green metal ruler", "blue fabric notebook", "wooden desk"], "negative_prompt": "blurred objects, low detail, mixed textures, chaotic composition", "diversity_tags": ["metal", "notebook", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1bf923ec2a92", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.6207, "diversity_score": 0.1793, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_1bf923ec2a92", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_1bf923ec2a92/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23825605, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_1bf923ec2a92/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_1bf923ec2a92/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_1bf923ec2a92/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1346829, "qwen_elapsed_seconds": 5.231, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_1bf923ec2a92/record.json"} +{"prompt_id": "occlusion_overlap_d2_2a07d97ed5d5", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic scene of a black metal key partially obscured by a blue fabric on a stone wall with high contrast boundaries.", "target_evidence": "black metal key", "expected_visual_elements": ["black metal key", "blue fabric", "stone wall"], "negative_prompt": "low resolution, distorted perspective, overlapping elements", "diversity_tags": ["metal", "fabric", "stone wall"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2a07d97ed5d5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_2a07d97ed5d5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_2a07d97ed5d5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23787448, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_2a07d97ed5d5/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_2a07d97ed5d5/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_2a07d97ed5d5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3509497, "qwen_elapsed_seconds": 5.249, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_2a07d97ed5d5/record.json"} +{"prompt_id": "occlusion_overlap_d2_390609673919", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic view of a white ceramic plate partially covered by a green leaf on a wooden counter with clean separation.", "target_evidence": "white ceramic plate", "expected_visual_elements": ["white ceramic plate", "green leaf", "wooden counter"], "negative_prompt": "low contrast, soft focus, overlapping textures", "diversity_tags": ["ceramic", "leaf", "wooden counter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "390609673919", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_390609673919", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_390609673919/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24794419, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_390609673919/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_390609673919/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_390609673919/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6760818, "qwen_elapsed_seconds": 5.185, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_390609673919/record.json"} +{"prompt_id": "occlusion_overlap_d2_45a650a475e5", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic view of a blue glass bottle partially covered by a white cloth draped over a metal shelf with sharp foreground separation.", "target_evidence": "blue glass bottle", "expected_visual_elements": ["blue glass bottle", "white cloth", "metal shelf"], "negative_prompt": "low resolution, distorted perspective, overlapping textures", "diversity_tags": ["glass", "cloth", "metal shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "45a650a475e5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5185, "diversity_score": 0.2815, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_45a650a475e5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_45a650a475e5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28766207, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_45a650a475e5/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_45a650a475e5/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_45a650a475e5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8418068, "qwen_elapsed_seconds": 5.215, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_45a650a475e5/record.json"} +{"prompt_id": "occlusion_overlap_d2_4d201414a1e3", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic setup of a red plastic container partially obscured by a black book on a tile floor with sharp foreground edges.", "target_evidence": "red plastic container", "expected_visual_elements": ["red plastic container", "black book", "tile floor"], "negative_prompt": "faded colors, low contrast, chaotic arrangement", "diversity_tags": ["plastic", "book", "tile floor"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4d201414a1e3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_4d201414a1e3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_4d201414a1e3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27754254, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_4d201414a1e3/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_4d201414a1e3/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_4d201414a1e3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5097810, "qwen_elapsed_seconds": 5.218, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_4d201414a1e3/record.json"} +{"prompt_id": "occlusion_overlap_d2_6c34e5c814eb", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic scene of a red ceramic mug partially hidden under a green fabric cloth on a wooden table, with clear foreground separation and sharp focus", "target_evidence": "red ceramic mug", "expected_visual_elements": ["red ceramic mug", "green fabric cloth", "wooden table"], "negative_prompt": "blurry edges, low contrast, chaotic composition, overlapping textures", "diversity_tags": ["ceramic", "fabric", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6c34e5c814eb", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.7586, "diversity_score": 0.0414, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_6c34e5c814eb", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_6c34e5c814eb/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24274362, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_6c34e5c814eb/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_6c34e5c814eb/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_6c34e5c814eb/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3610987, "qwen_elapsed_seconds": 5.219, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_6c34e5c814eb/record.json"} +{"prompt_id": "occlusion_overlap_d2_736a639062c2", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic view of a blue glass bottle partially covered by a transparent glass sheet on a steel countertop, with high contrast and clean boundaries", "target_evidence": "blue glass bottle", "expected_visual_elements": ["blue glass bottle", "transparent glass sheet", "steel countertop"], "negative_prompt": "low detail, poor depth, smudged edges, background clutter", "diversity_tags": ["glass", "transparent", "countertop"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "736a639062c2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5517, "diversity_score": 0.2483, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_736a639062c2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_736a639062c2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23366857, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_736a639062c2/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_736a639062c2/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_736a639062c2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1437497, "qwen_elapsed_seconds": 5.206, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_736a639062c2/record.json"} +{"prompt_id": "occlusion_overlap_d2_771a8818b447", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic view of a red ceramic mug partially hidden under a green velvet cloth on a wooden table, with clear foreground separation and sharp focus on the mug's handle", "target_evidence": "red ceramic mug", "expected_visual_elements": ["red ceramic mug", "green velvet cloth", "wooden table"], "negative_prompt": "blurry edges, low contrast, chaotic arrangement, overlapping textures, unrealistic lighting", "diversity_tags": ["ceramic", "velvet", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "771a8818b447", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.7586, "diversity_score": 0.0414, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_771a8818b447", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_771a8818b447/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21697775, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_771a8818b447/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_771a8818b447/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_771a8818b447/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2207926, "qwen_elapsed_seconds": 5.227, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_771a8818b447/record.json"} +{"prompt_id": "occlusion_overlap_d2_990a4f93e3da", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic scene of a silver metal spoon partially overlapped by a transparent glass plate on a marble countertop, with clean separation and high clarity", "target_evidence": "silver metal spoon", "expected_visual_elements": ["silver metal spoon", "transparent glass plate", "marble countertop"], "negative_prompt": "low detail, smudged edges, poor depth, chaotic composition", "diversity_tags": ["metal", "transparent", "countertop"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "990a4f93e3da", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.6, "diversity_score": 0.2, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_990a4f93e3da", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_990a4f93e3da/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24805009, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_990a4f93e3da/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_990a4f93e3da/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_990a4f93e3da/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 29141, "qwen_elapsed_seconds": 5.228, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_990a4f93e3da/record.json"} +{"prompt_id": "occlusion_overlap_d2_9ec5f87cf461", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic setup of a blue metal pen partially covered by a transparent glass cup on a steel desk, with high contrast and clear separation", "target_evidence": "blue metal pen", "expected_visual_elements": ["blue metal pen", "transparent glass cup", "steel desk"], "negative_prompt": "low resolution, poor lighting, merged objects, low detail", "diversity_tags": ["metal", "transparent", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "9ec5f87cf461", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.5517, "diversity_score": 0.2483, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_9ec5f87cf461", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_9ec5f87cf461/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22072790, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_9ec5f87cf461/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_9ec5f87cf461/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_9ec5f87cf461/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 916676, "qwen_elapsed_seconds": 5.223, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_9ec5f87cf461/record.json"} +{"prompt_id": "occlusion_overlap_d2_a184bb94fa5c", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic scene of a yellow plastic ruler partially overlapped by a white fabric cloth on a concrete floor, with clean boundaries and high clarity", "target_evidence": "yellow plastic ruler", "expected_visual_elements": ["yellow plastic ruler", "white fabric cloth", "concrete floor"], "negative_prompt": "low detail, smudged edges, poor depth, chaotic composition", "diversity_tags": ["plastic", "fabric", "floor"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a184bb94fa5c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.5172, "diversity_score": 0.2828, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_a184bb94fa5c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_a184bb94fa5c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21470185, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_a184bb94fa5c/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_a184bb94fa5c/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_a184bb94fa5c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2741108, "qwen_elapsed_seconds": 5.168, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_a184bb94fa5c/record.json"} +{"prompt_id": "occlusion_overlap_d2_ac4b28df9349", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic setup of a white paper cup partially hidden under a brown envelope on a plastic desk with clear depth layers.", "target_evidence": "white paper cup", "expected_visual_elements": ["white paper cup", "brown envelope", "plastic desk"], "negative_prompt": "faded colors, low contrast, soft shadows", "diversity_tags": ["paper", "envelope", "plastic desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ac4b28df9349", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4286, "diversity_score": 0.3714, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_ac4b28df9349", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_ac4b28df9349/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27506397, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_ac4b28df9349/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_ac4b28df9349/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_ac4b28df9349/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1416580, "qwen_elapsed_seconds": 5.234, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_ac4b28df9349/record.json"} +{"prompt_id": "occlusion_overlap_d2_b9c7d07a7b71", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic scene of a red ceramic mug partially hidden under a green notebook on a wooden table with clear depth separation.", "target_evidence": "red ceramic mug", "expected_visual_elements": ["red ceramic mug", "green notebook", "wooden table"], "negative_prompt": "blurry, low contrast, chaotic composition, overlapping elements", "diversity_tags": ["ceramic", "notebook", "wooden table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b9c7d07a7b71", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.6923, "diversity_score": 0.1077, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_b9c7d07a7b71", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_b9c7d07a7b71/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21948347, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_b9c7d07a7b71/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_b9c7d07a7b71/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_b9c7d07a7b71/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3828340, "qwen_elapsed_seconds": 5.248, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_b9c7d07a7b71/record.json"} +{"prompt_id": "occlusion_overlap_d2_c01ec8540ecc", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic setup of a yellow metal key partially obscured by a black leather wallet on a white tile floor, with clear depth ordering and sharp focus", "target_evidence": "yellow metal key", "expected_visual_elements": ["yellow metal key", "black leather wallet", "white tile floor"], "negative_prompt": "low resolution, poor lighting, merged objects, low contrast", "diversity_tags": ["metal", "leather", "floor"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c01ec8540ecc", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4839, "diversity_score": 0.3161, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_c01ec8540ecc", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_c01ec8540ecc/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27521503, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_c01ec8540ecc/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_c01ec8540ecc/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_c01ec8540ecc/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2317211, "qwen_elapsed_seconds": 5.236, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_c01ec8540ecc/record.json"} +{"prompt_id": "occlusion_overlap_d2_c37675a0d2e2", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic scene with a yellow metal bowl partially behind a transparent glass on a glass countertop with clear depth separation.", "target_evidence": "yellow metal bowl", "expected_visual_elements": ["yellow metal bowl", "transparent glass", "glass countertop"], "negative_prompt": "low detail, soft focus, overlapping textures", "diversity_tags": ["metal", "glass", "countertop"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c37675a0d2e2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.5455, "diversity_score": 0.2545, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_c37675a0d2e2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_c37675a0d2e2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27265268, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_c37675a0d2e2/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_c37675a0d2e2/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_c37675a0d2e2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6887837, "qwen_elapsed_seconds": 5.18, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_c37675a0d2e2/record.json"} +{"prompt_id": "occlusion_overlap_d2_e9f5b3fc2a22", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic setup of a yellow plastic chair with a black laptop partially obscuring it on a tile floor with high contrast edges.", "target_evidence": "yellow plastic chair", "expected_visual_elements": ["yellow plastic chair", "black laptop", "tile floor"], "negative_prompt": "faded colors, soft shadows, overlapping layers", "diversity_tags": ["plastic", "laptop", "tile floor"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e9f5b3fc2a22", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_e9f5b3fc2a22", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_e9f5b3fc2a22/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28736291, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_e9f5b3fc2a22/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_e9f5b3fc2a22/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_e9f5b3fc2a22/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2703492, "qwen_elapsed_seconds": 5.21, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_e9f5b3fc2a22/record.json"} +{"prompt_id": "occlusion_overlap_d2_ef1916ec63f9", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic scene with a silver spoon partially behind a transparent glass of water on a glass countertop with clean separation.", "target_evidence": "silver spoon", "expected_visual_elements": ["silver spoon", "transparent glass", "glass countertop"], "negative_prompt": "low detail, unclear boundaries, soft focus", "diversity_tags": ["spoon", "glass", "countertop"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ef1916ec63f9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.6, "diversity_score": 0.2, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_ef1916ec63f9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_ef1916ec63f9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21488922, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_ef1916ec63f9/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_ef1916ec63f9/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_ef1916ec63f9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1243107, "qwen_elapsed_seconds": 5.162, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_ef1916ec63f9/record.json"} +{"prompt_id": "occlusion_overlap_d2_efb5d9f55543", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 2, "image_prompt": "synthetic photorealistic scene with a white plastic cup partially overlapped by a green plastic container on a concrete wall, with high contrast and clean separation", "target_evidence": "white plastic cup", "expected_visual_elements": ["white plastic cup", "green plastic container", "concrete wall"], "negative_prompt": "low clarity, messy composition, overlapping textures, poor depth", "diversity_tags": ["plastic", "container", "wall"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "efb5d9f55543", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.5172, "diversity_score": 0.2828, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d2_efb5d9f55543", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d2_efb5d9f55543/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24269277, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d2_efb5d9f55543/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_efb5d9f55543/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_efb5d9f55543/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7261122, "qwen_elapsed_seconds": 5.212, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d2_efb5d9f55543/record.json"} +{"prompt_id": "occlusion_overlap_d3_0ce9a3f2833c", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "Synthetic photorealistic view of a blue metal tool partially overlapping with a transparent plastic container on a dark wooden shelf, with clear separation between the tool and container.", "target_evidence": "blue metal tool", "expected_visual_elements": ["blue metal tool", "transparent plastic container", "dark wooden shelf"], "negative_prompt": "low resolution, low contrast, cluttered background, poor lighting", "diversity_tags": ["metal", "plastic", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0ce9a3f2833c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5667, "diversity_score": 0.2333, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_0ce9a3f2833c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_0ce9a3f2833c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22437835, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_0ce9a3f2833c/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_0ce9a3f2833c/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_0ce9a3f2833c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7952245, "qwen_elapsed_seconds": 5.188, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_0ce9a3f2833c/record.json"} +{"prompt_id": "occlusion_overlap_d3_0d616f7d2bdd", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "Synthetic photorealistic scene of a yellow plastic container partially overlapping with a transparent glass bottle on a light-colored floor, with clear depth ordering.", "target_evidence": "yellow plastic container", "expected_visual_elements": ["yellow plastic container", "transparent glass bottle", "light-colored floor"], "negative_prompt": "low detail, poor depth, overlapping textures, blurry edges", "diversity_tags": ["plastic", "glass", "floor"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0d616f7d2bdd", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_0d616f7d2bdd", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_0d616f7d2bdd/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25969429, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_0d616f7d2bdd/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_0d616f7d2bdd/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_0d616f7d2bdd/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1464215, "qwen_elapsed_seconds": 5.204, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_0d616f7d2bdd/record.json"} +{"prompt_id": "occlusion_overlap_d3_229260b847eb", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "synthetic photorealistic scene of a red ceramic mug partially hidden under a blue stack of books on a wooden table, clean boundaries and sharp focus for clear separation", "target_evidence": "red ceramic mug", "expected_visual_elements": ["red ceramic mug", "blue stack of books", "wooden table"], "negative_prompt": "blurry, low contrast, overlapping textures, chaotic composition", "diversity_tags": ["ceramic", "books", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "229260b847eb", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.6786, "diversity_score": 0.1214, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_229260b847eb", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_229260b847eb/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29604978, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_229260b847eb/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_229260b847eb/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_229260b847eb/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4771200, "qwen_elapsed_seconds": 5.212, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_229260b847eb/record.json"} +{"prompt_id": "occlusion_overlap_d3_38814fd62e99", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "Synthetic photorealistic scene of a green ceramic mug partially covered by a black leather strap on a dark wooden desk, with clear boundaries between the mug and strap.", "target_evidence": "green ceramic mug", "expected_visual_elements": ["green ceramic mug", "black leather strap", "dark wooden desk"], "negative_prompt": "low resolution, low contrast, cluttered background, poor lighting", "diversity_tags": ["ceramic", "leather", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "38814fd62e99", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4643, "diversity_score": 0.3357, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_38814fd62e99", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_38814fd62e99/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28960951, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_38814fd62e99/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_38814fd62e99/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_38814fd62e99/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 147238, "qwen_elapsed_seconds": 5.225, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_38814fd62e99/record.json"} +{"prompt_id": "occlusion_overlap_d3_399299299133", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "Synthetic photorealistic view of a blue ceramic mug partially hidden behind a stack of white books on a dark wooden shelf, with clear separation between the mug and books.", "target_evidence": "blue ceramic mug", "expected_visual_elements": ["blue ceramic mug", "white books", "dark wooden shelf"], "negative_prompt": "low detail, poor depth, overlapping textures, blurry edges", "diversity_tags": ["ceramic", "books", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "399299299133", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5667, "diversity_score": 0.2333, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_399299299133", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_399299299133/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22245849, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_399299299133/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_399299299133/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_399299299133/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8336391, "qwen_elapsed_seconds": 5.28, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_399299299133/record.json"} +{"prompt_id": "occlusion_overlap_d3_52fd8c4363d0", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "synthetic photorealistic view of a blue metal key partially hidden under a brown leather wallet on a stone surface, clear depth for layer separation", "target_evidence": "blue metal key", "expected_visual_elements": ["blue metal key", "brown leather wallet", "stone surface"], "negative_prompt": "faded colors, low resolution, unclear boundaries, chaotic layout", "diversity_tags": ["metal", "leather", "stone"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "52fd8c4363d0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4286, "diversity_score": 0.3714, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_52fd8c4363d0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_52fd8c4363d0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29068148, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_52fd8c4363d0/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_52fd8c4363d0/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_52fd8c4363d0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7447282, "qwen_elapsed_seconds": 5.242, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_52fd8c4363d0/record.json"} +{"prompt_id": "occlusion_overlap_d3_54165ab16342", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "synthetic photorealistic setup of a white paper notebook partially covered by a black ink pen on a glass desk, high contrast for visual separability", "target_evidence": "white paper notebook", "expected_visual_elements": ["white paper notebook", "black ink pen", "glass desk"], "negative_prompt": "overexposed, low detail, blurred edges, unclear depth", "diversity_tags": ["paper", "ink", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "54165ab16342", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4516, "diversity_score": 0.3484, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_54165ab16342", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_54165ab16342/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24012048, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_54165ab16342/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_54165ab16342/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_54165ab16342/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 505874, "qwen_elapsed_seconds": 5.145, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_54165ab16342/record.json"} +{"prompt_id": "occlusion_overlap_d3_5af081ababf8", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "Synthetic photorealistic view of a green wooden box partially obscured by a white fabric cloth on a gray wall, with clean boundaries between the box and cloth.", "target_evidence": "green wooden box", "expected_visual_elements": ["green wooden box", "white fabric cloth", "gray wall"], "negative_prompt": "low contrast, poor depth, overlapping textures, blurry edges", "diversity_tags": ["wooden", "fabric", "wall"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5af081ababf8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4643, "diversity_score": 0.3357, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_5af081ababf8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_5af081ababf8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24343787, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_5af081ababf8/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_5af081ababf8/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_5af081ababf8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4559584, "qwen_elapsed_seconds": 5.239, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_5af081ababf8/record.json"} +{"prompt_id": "occlusion_overlap_d3_676686ef2763", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "synthetic photorealistic setup of a blue glass marble partially covered by a red fabric cloth on a tile floor, sharp focus to ensure clean separation", "target_evidence": "blue glass marble", "expected_visual_elements": ["blue glass marble", "red fabric cloth", "tile floor"], "negative_prompt": "overexposed, low resolution, blurred edges, chaotic layout", "diversity_tags": ["glass", "fabric", "floor"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "676686ef2763", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.5357, "diversity_score": 0.2643, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_676686ef2763", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_676686ef2763/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28141319, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_676686ef2763/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_676686ef2763/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_676686ef2763/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7661383, "qwen_elapsed_seconds": 5.246, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_676686ef2763/record.json"} +{"prompt_id": "occlusion_overlap_d3_7ffd6596c928", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "Synthetic photorealistic setup of a silver key partially covered by a black leather wallet on a white countertop, with high contrast between the key and wallet.", "target_evidence": "silver key", "expected_visual_elements": ["silver key", "black leather wallet", "white countertop"], "negative_prompt": "low resolution, low contrast, cluttered background, poor lighting", "diversity_tags": ["key", "wallet", "countertop"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7ffd6596c928", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_7ffd6596c928", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_7ffd6596c928/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25884626, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_7ffd6596c928/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_7ffd6596c928/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_7ffd6596c928/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2911840, "qwen_elapsed_seconds": 5.197, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_7ffd6596c928/record.json"} +{"prompt_id": "occlusion_overlap_d3_80aaedffd4b3", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "synthetic photorealistic view of a green glass vial overlapping with a transparent plastic container on a metal shelf, high contrast for visual separability", "target_evidence": "green glass vial", "expected_visual_elements": ["green glass vial", "transparent plastic container", "metal shelf"], "negative_prompt": "low resolution, unclear depth, messy lighting, distorted perspective", "diversity_tags": ["glass", "plastic", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "80aaedffd4b3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_80aaedffd4b3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_80aaedffd4b3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23465776, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_80aaedffd4b3/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_80aaedffd4b3/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_80aaedffd4b3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9240909, "qwen_elapsed_seconds": 5.26, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_80aaedffd4b3/record.json"} +{"prompt_id": "occlusion_overlap_d3_849dbb5e8af5", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "Synthetic photorealistic scene of a white plastic cup partially covered by a green sponge on a light-colored floor, with sharp focus on the cup's rim.", "target_evidence": "white plastic cup", "expected_visual_elements": ["white plastic cup", "green sponge", "light-colored floor"], "negative_prompt": "low detail, poor depth, overlapping textures, blurry edges", "diversity_tags": ["plastic", "sponge", "floor"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "849dbb5e8af5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_849dbb5e8af5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_849dbb5e8af5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24802859, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_849dbb5e8af5/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_849dbb5e8af5/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_849dbb5e8af5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8387781, "qwen_elapsed_seconds": 5.22, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_849dbb5e8af5/record.json"} +{"prompt_id": "occlusion_overlap_d3_857374576ffc", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "Synthetic photorealistic setup of a red rubber ball partially hidden behind a white paper towel on a light-colored countertop, with high contrast between the ball and towel.", "target_evidence": "red rubber ball", "expected_visual_elements": ["red rubber ball", "white paper towel", "light-colored countertop"], "negative_prompt": "low detail, poor depth, overlapping textures, blurry edges", "diversity_tags": ["rubber", "paper", "countertop"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "857374576ffc", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4839, "diversity_score": 0.3161, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_857374576ffc", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_857374576ffc/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27330131, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_857374576ffc/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_857374576ffc/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_857374576ffc/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5780099, "qwen_elapsed_seconds": 5.235, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_857374576ffc/record.json"} +{"prompt_id": "occlusion_overlap_d3_c03fc3b263bf", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "synthetic photorealistic scene with a white plastic cup partially occluded by a green plant on a tile floor, sharp focus to ensure clean separation", "target_evidence": "white plastic cup", "expected_visual_elements": ["white plastic cup", "green plant", "tile floor"], "negative_prompt": "low detail, washed out colors, overlapping shadows, unclear layering", "diversity_tags": ["plastic", "plant", "floor"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c03fc3b263bf", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_c03fc3b263bf", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_c03fc3b263bf/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25744069, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_c03fc3b263bf/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_c03fc3b263bf/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_c03fc3b263bf/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6945136, "qwen_elapsed_seconds": 5.224, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_c03fc3b263bf/record.json"} +{"prompt_id": "occlusion_overlap_d3_c3936cbdd3f3", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "synthetic photorealistic scene of a silver wireframe cube overlapping with a transparent acrylic box on a concrete wall, clear foreground separation", "target_evidence": "silver wireframe cube", "expected_visual_elements": ["silver wireframe cube", "transparent acrylic box", "concrete wall"], "negative_prompt": "low resolution, unclear depth, disordered lighting, distorted perspective", "diversity_tags": ["wireframe", "acrylic", "wall"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c3936cbdd3f3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4, "diversity_score": 0.4, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_c3936cbdd3f3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_c3936cbdd3f3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28911599, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_c3936cbdd3f3/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_c3936cbdd3f3/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_c3936cbdd3f3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9556686, "qwen_elapsed_seconds": 5.136, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_c3936cbdd3f3/record.json"} +{"prompt_id": "occlusion_overlap_d3_c6693edc51ab", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "synthetic photorealistic setup of a yellow rubber ball partially covered by a black fabric cloth draped over a wooden counter, visible fraction controlled for separation testing", "target_evidence": "yellow rubber ball", "expected_visual_elements": ["yellow rubber ball", "black fabric cloth", "wooden counter"], "negative_prompt": "overexposed, out of focus, disordered arrangement, unclear edges", "diversity_tags": ["rubber", "fabric", "counter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c6693edc51ab", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.3939, "diversity_score": 0.4061, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_c6693edc51ab", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_c6693edc51ab/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26544552, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_c6693edc51ab/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_c6693edc51ab/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_c6693edc51ab/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 778905, "qwen_elapsed_seconds": 5.225, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_c6693edc51ab/record.json"} +{"prompt_id": "occlusion_overlap_d3_cd090365e8d2", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "Synthetic photorealistic scene of a red glass jar partially covered by a green leafy plant on a wooden table, with sharp focus on the jar's neck.", "target_evidence": "red glass jar", "expected_visual_elements": ["red glass jar", "green leafy plant", "wooden table"], "negative_prompt": "blurry background, low contrast, chaotic arrangement, overlapping textures", "diversity_tags": ["glass", "plant", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cd090365e8d2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.7037, "diversity_score": 0.0963, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_cd090365e8d2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_cd090365e8d2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20568373, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_cd090365e8d2/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_cd090365e8d2/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_cd090365e8d2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8217597, "qwen_elapsed_seconds": 5.283, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_cd090365e8d2/record.json"} +{"prompt_id": "occlusion_overlap_d3_ced9b0d35e62", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "Synthetic photorealistic setup of a black metal spoon partially hidden behind a white ceramic plate on a dark wooden counter, with clear foreground separation.", "target_evidence": "black metal spoon", "expected_visual_elements": ["black metal spoon", "white ceramic plate", "dark wooden counter"], "negative_prompt": "low resolution, low contrast, cluttered background, poor lighting", "diversity_tags": ["metal", "ceramic", "counter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ced9b0d35e62", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.5556, "diversity_score": 0.2444, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_ced9b0d35e62", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_ced9b0d35e62/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26110365, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_ced9b0d35e62/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_ced9b0d35e62/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_ced9b0d35e62/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4999655, "qwen_elapsed_seconds": 5.229, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_ced9b0d35e62/record.json"} +{"prompt_id": "occlusion_overlap_d3_d222b0f5c9b3", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "synthetic photorealistic scene with a black metal ruler partially hidden under a white foam board on a wooden counter, clean boundaries for clear separation", "target_evidence": "black metal ruler", "expected_visual_elements": ["black metal ruler", "white foam board", "wooden counter"], "negative_prompt": "low detail, washed out colors, overlapping shadows, unclear layering", "diversity_tags": ["metal", "foam", "counter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d222b0f5c9b3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.4667, "diversity_score": 0.3333, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_d222b0f5c9b3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_d222b0f5c9b3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25954142, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_d222b0f5c9b3/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_d222b0f5c9b3/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_d222b0f5c9b3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3891297, "qwen_elapsed_seconds": 5.179, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_d222b0f5c9b3/record.json"} +{"prompt_id": "occlusion_overlap_d3_fe20fc3947fb", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 3, "image_prompt": "synthetic photorealistic view of a pink foam block partially occluded by a green leaf on a wooden shelf, visible fraction controlled for separation testing", "target_evidence": "pink foam block", "expected_visual_elements": ["pink foam block", "green leaf", "wooden shelf"], "negative_prompt": "low contrast, out of focus, disordered arrangement, unclear edges", "diversity_tags": ["foam", "leaf", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "fe20fc3947fb", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4138, "diversity_score": 0.3862, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d3_fe20fc3947fb", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d3_fe20fc3947fb/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21749024, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d3_fe20fc3947fb/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_fe20fc3947fb/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_fe20fc3947fb/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 571612, "qwen_elapsed_seconds": 5.187, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d3_fe20fc3947fb/record.json"} +{"prompt_id": "occlusion_overlap_d4_164d94066bd2", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic scene of a yellow plastic container partially hidden behind a green fabric pouch on a light-colored shelf with clean edges", "target_evidence": "yellow plastic container", "expected_visual_elements": ["yellow plastic container", "green fabric pouch"], "negative_prompt": "low detail, motion blur, overlapping textures", "diversity_tags": ["plastic", "fabric", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "164d94066bd2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4483, "diversity_score": 0.3517, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_164d94066bd2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_164d94066bd2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29805271, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_164d94066bd2/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_164d94066bd2/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_164d94066bd2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9327819, "qwen_elapsed_seconds": 5.203, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_164d94066bd2/record.json"} +{"prompt_id": "occlusion_overlap_d4_1851da3abc78", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic setup of a white ceramic plate partially covered by a dark green leaf, with sharp focus and clear depth", "target_evidence": "white ceramic plate", "expected_visual_elements": ["white ceramic plate", "dark green leaf"], "negative_prompt": "low detail, blurred edges, overlapping textures, chaotic layout", "diversity_tags": ["ceramic", "plant", "color contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1851da3abc78", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.5185, "diversity_score": 0.2815, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_1851da3abc78", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_1851da3abc78/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23209465, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_1851da3abc78/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_1851da3abc78/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_1851da3abc78/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2534987, "qwen_elapsed_seconds": 5.161, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_1851da3abc78/record.json"} +{"prompt_id": "occlusion_overlap_d4_1f08a12f86f1", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a red rubber duck partially covered by a blue fabric curtain, with sharp focus and clean boundaries", "target_evidence": "red rubber duck", "expected_visual_elements": ["red rubber duck", "blue fabric curtain"], "negative_prompt": "low detail, unclear depth, blurry edges, overlapping textures", "diversity_tags": ["rubber", "fabric", "color contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1f08a12f86f1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_1f08a12f86f1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_1f08a12f86f1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22265928, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_1f08a12f86f1/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_1f08a12f86f1/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_1f08a12f86f1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4986567, "qwen_elapsed_seconds": 5.224, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_1f08a12f86f1/record.json"} +{"prompt_id": "occlusion_overlap_d4_220a39f4246d", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic setup of a black metal plate partially obscured by a red fabric scarf on a dark wooden table with clear depth", "target_evidence": "black metal plate", "expected_visual_elements": ["black metal plate", "red fabric scarf"], "negative_prompt": "blurry, low contrast, chaotic layout", "diversity_tags": ["metal", "fabric", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "220a39f4246d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4828, "diversity_score": 0.3172, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_220a39f4246d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_220a39f4246d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27525521, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_220a39f4246d/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_220a39f4246d/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_220a39f4246d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 394043, "qwen_elapsed_seconds": 5.145, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_220a39f4246d/record.json"} +{"prompt_id": "occlusion_overlap_d4_258bbe19ecf4", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a red ceramic mug partially covered by a green leafy plant on a wooden table with sharp focus on the mug's handle", "target_evidence": "red ceramic mug", "expected_visual_elements": ["red ceramic mug", "green leafy plant"], "negative_prompt": "blurry, low contrast, chaotic background, overlapping textures", "diversity_tags": ["ceramic", "plant", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "258bbe19ecf4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.7037, "diversity_score": 0.0963, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_258bbe19ecf4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_258bbe19ecf4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26448452, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_258bbe19ecf4/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_258bbe19ecf4/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_258bbe19ecf4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1351844, "qwen_elapsed_seconds": 5.242, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_258bbe19ecf4/record.json"} +{"prompt_id": "occlusion_overlap_d4_2b0b98cdfd96", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic tabletop with a white porcelain teacup partially covered by a blue fabric cloth, showing clear separation and focus", "target_evidence": "white porcelain teacup", "expected_visual_elements": ["white porcelain teacup", "blue fabric cloth"], "negative_prompt": "blurry edges, low contrast, overlapping textures, chaotic layout", "diversity_tags": ["porcelain", "fabric", "color contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2b0b98cdfd96", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tabletop with a", "max_similarity_to_other_prompt": 0.3929, "diversity_score": 0.4071, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_2b0b98cdfd96", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_2b0b98cdfd96/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28253662, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_2b0b98cdfd96/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_2b0b98cdfd96/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_2b0b98cdfd96/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8576515, "qwen_elapsed_seconds": 5.165, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_2b0b98cdfd96/record.json"} +{"prompt_id": "occlusion_overlap_d4_419d52ce8d7f", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic setup of a blue rubber ball partially obscured by a wooden chair leg, with sharp focus on the visible ball portion", "target_evidence": "blue rubber ball", "expected_visual_elements": ["blue rubber ball", "wooden chair leg"], "negative_prompt": "overlapping shadows, low contrast, blurred edges, low detail", "diversity_tags": ["rubber", "wood", "occlusion"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "419d52ce8d7f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4138, "diversity_score": 0.3862, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_419d52ce8d7f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_419d52ce8d7f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28227640, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_419d52ce8d7f/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_419d52ce8d7f/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_419d52ce8d7f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3790556, "qwen_elapsed_seconds": 5.213, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_419d52ce8d7f/record.json"} +{"prompt_id": "occlusion_overlap_d4_478463ffdd5f", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a purple ceramic vase partially overlapped by a transparent plastic sheet on a light gray desk with high contrast", "target_evidence": "purple ceramic vase", "expected_visual_elements": ["purple ceramic vase", "transparent plastic sheet"], "negative_prompt": "low resolution, color distortion, unclear edges", "diversity_tags": ["ceramic", "plastic", "desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "478463ffdd5f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4483, "diversity_score": 0.3517, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_478463ffdd5f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_478463ffdd5f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21871924, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_478463ffdd5f/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_478463ffdd5f/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_478463ffdd5f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7379789, "qwen_elapsed_seconds": 5.159, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_478463ffdd5f/record.json"} +{"prompt_id": "occlusion_overlap_d4_4d8aed4b9387", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a white plastic cup partially overlapped by a transparent glass jar on a light-colored countertop with clear boundaries", "target_evidence": "white plastic cup", "expected_visual_elements": ["white plastic cup", "transparent glass jar"], "negative_prompt": "low resolution, color bleed, unclear edges", "diversity_tags": ["plastic", "glass", "counter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4d8aed4b9387", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4667, "diversity_score": 0.3333, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_4d8aed4b9387", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_4d8aed4b9387/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29716249, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_4d8aed4b9387/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_4d8aed4b9387/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_4d8aed4b9387/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4188131, "qwen_elapsed_seconds": 5.252, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_4d8aed4b9387/record.json"} +{"prompt_id": "occlusion_overlap_d4_5d46f33e1739", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a white ceramic bowl partially covered by a transparent glass lid on a wooden counter with sharp focus and clear boundaries", "target_evidence": "white ceramic bowl", "expected_visual_elements": ["white ceramic bowl", "transparent glass lid"], "negative_prompt": "low resolution, color bleed, unclear edges", "diversity_tags": ["ceramic", "glass", "counter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5d46f33e1739", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_5d46f33e1739", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_5d46f33e1739/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26374362, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_5d46f33e1739/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_5d46f33e1739/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_5d46f33e1739/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5581713, "qwen_elapsed_seconds": 5.225, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_5d46f33e1739/record.json"} +{"prompt_id": "occlusion_overlap_d4_907a3ed86086", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic scene of a yellow plastic cup partially overlapped by a black transparent sheet, showing clean separation and focus", "target_evidence": "yellow plastic cup", "expected_visual_elements": ["yellow plastic cup", "black transparent sheet"], "negative_prompt": "low contrast, blurry edges, overlapping textures, chaotic layout", "diversity_tags": ["plastic", "transparent", "overlap"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "907a3ed86086", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4444, "diversity_score": 0.3556, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_907a3ed86086", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_907a3ed86086/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23278429, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_907a3ed86086/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_907a3ed86086/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_907a3ed86086/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2246100, "qwen_elapsed_seconds": 5.211, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_907a3ed86086/record.json"} +{"prompt_id": "occlusion_overlap_d4_c8eaafde938f", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic scene of a silver metal spoon partially hidden behind a brown cardboard box on a wooden floor with sharp focus", "target_evidence": "silver metal spoon", "expected_visual_elements": ["silver metal spoon", "brown cardboard box"], "negative_prompt": "low detail, motion blur, unclear edges", "diversity_tags": ["metal", "cardboard", "floor"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c8eaafde938f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4231, "diversity_score": 0.3769, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_c8eaafde938f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_c8eaafde938f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21600157, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_c8eaafde938f/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_c8eaafde938f/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_c8eaafde938f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2602907, "qwen_elapsed_seconds": 5.129, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_c8eaafde938f/record.json"} +{"prompt_id": "occlusion_overlap_d4_cb51e79a0b45", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic scene of a yellow metal key partially obscured by a black leather wallet on a dark wooden shelf with high contrast", "target_evidence": "yellow metal key", "expected_visual_elements": ["yellow metal key", "black leather wallet"], "negative_prompt": "low detail, cluttered, overlapping edges", "diversity_tags": ["metal", "wallet", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cb51e79a0b45", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_cb51e79a0b45", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_cb51e79a0b45/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24107749, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_cb51e79a0b45/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_cb51e79a0b45/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_cb51e79a0b45/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 449431, "qwen_elapsed_seconds": 5.156, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_cb51e79a0b45/record.json"} +{"prompt_id": "occlusion_overlap_d4_ccc28ce59ecc", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a glass vial partially covered by a woven basket on a wooden table, with clear foreground separation and high contrast between transparent and opaque materials", "target_evidence": "glass vial", "expected_visual_elements": ["glass vial", "woven basket", "wooden table"], "negative_prompt": "blurry edges, low contrast, overlapping textures, chaotic composition", "diversity_tags": ["transparent overlap", "material contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ccc28ce59ecc", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5161, "diversity_score": 0.2839, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_ccc28ce59ecc", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_ccc28ce59ecc/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22504270, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_ccc28ce59ecc/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_ccc28ce59ecc/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_ccc28ce59ecc/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 164581, "qwen_elapsed_seconds": 5.178, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_ccc28ce59ecc/record.json"} +{"prompt_id": "occlusion_overlap_d4_ce67d780f583", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a brown leather wallet partially hidden under a transparent plastic bag, with clear foreground and background", "target_evidence": "brown leather wallet", "expected_visual_elements": ["brown leather wallet", "transparent plastic bag"], "negative_prompt": "low contrast, blurry edges, overlapping textures, chaotic layout", "diversity_tags": ["leather", "plastic", "transparent"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ce67d780f583", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4483, "diversity_score": 0.3517, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_ce67d780f583", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_ce67d780f583/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27695142, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_ce67d780f583/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_ce67d780f583/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_ce67d780f583/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2017793, "qwen_elapsed_seconds": 5.193, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_ce67d780f583/record.json"} +{"prompt_id": "occlusion_overlap_d4_d8697951c02b", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic scene of a green metal file partially overlapping a wooden ruler, with sharp focus and clean boundaries", "target_evidence": "green metal file", "expected_visual_elements": ["green metal file", "wooden ruler"], "negative_prompt": "low detail, unclear depth, blurred edges, overlapping textures", "diversity_tags": ["metal", "wood", "overlap"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d8697951c02b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_d8697951c02b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_d8697951c02b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22350790, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_d8697951c02b/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_d8697951c02b/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_d8697951c02b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3703781, "qwen_elapsed_seconds": 5.295, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_d8697951c02b/record.json"} +{"prompt_id": "occlusion_overlap_d4_e04cbf7caec8", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic setup of a blue glass bottle partially hidden behind a stack of white paper on a counter with clear foreground separation", "target_evidence": "blue glass bottle", "expected_visual_elements": ["blue glass bottle", "white paper stack"], "negative_prompt": "low resolution, out-of-focus, distorted perspective", "diversity_tags": ["glass", "paper", "counter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e04cbf7caec8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.5556, "diversity_score": 0.2444, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_e04cbf7caec8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_e04cbf7caec8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20750814, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_e04cbf7caec8/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_e04cbf7caec8/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_e04cbf7caec8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4708876, "qwen_elapsed_seconds": 5.243, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_e04cbf7caec8/record.json"} +{"prompt_id": "occlusion_overlap_d4_e49197e6afab", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic scene of a red ceramic mug partially covered by a green leafy plant, with clear foreground separation and sharp focus", "target_evidence": "red ceramic mug", "expected_visual_elements": ["red ceramic mug", "green leafy plant"], "negative_prompt": "blurry background, low contrast, overlapping textures, chaotic layout", "diversity_tags": ["ceramic", "plant", "color contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e49197e6afab", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.6071, "diversity_score": 0.1929, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_e49197e6afab", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_e49197e6afab/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22614941, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_e49197e6afab/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_e49197e6afab/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_e49197e6afab/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5670801, "qwen_elapsed_seconds": 5.205, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_e49197e6afab/record.json"} +{"prompt_id": "occlusion_overlap_d4_f678c98454ac", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a silver key lying beneath a transparent glass jar filled with water, showing clear depth and material distinction", "target_evidence": "silver key", "expected_visual_elements": ["silver key", "transparent glass jar"], "negative_prompt": "low resolution, unclear boundaries, water reflections, out-of-focus", "diversity_tags": ["metal", "glass", "transparent"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f678c98454ac", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.3448, "diversity_score": 0.4552, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_f678c98454ac", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_f678c98454ac/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26774774, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_f678c98454ac/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_f678c98454ac/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_f678c98454ac/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2071535, "qwen_elapsed_seconds": 5.246, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_f678c98454ac/record.json"} +{"prompt_id": "occlusion_overlap_d4_ff52b6ba64cf", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 4, "image_prompt": "synthetic photorealistic scene with a yellow plastic container partially hidden under a black notebook, showing clean boundary and depth", "target_evidence": "yellow plastic container", "expected_visual_elements": ["yellow plastic container", "black notebook"], "negative_prompt": "low detail, chaotic arrangement, unclear layering, low contrast", "diversity_tags": ["plastic", "notebook", "color contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ff52b6ba64cf", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene with a", "max_similarity_to_other_prompt": 0.4231, "diversity_score": 0.3769, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d4_ff52b6ba64cf", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d4_ff52b6ba64cf/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25806292, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d4_ff52b6ba64cf/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_ff52b6ba64cf/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_ff52b6ba64cf/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4238405, "qwen_elapsed_seconds": 5.133, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d4_ff52b6ba64cf/record.json"} +{"prompt_id": "occlusion_overlap_d5_037b1ca0a450", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic setup of a yellow metal key partially hidden under a white ceramic plate, clean separation between objects", "target_evidence": "yellow metal key", "expected_visual_elements": ["yellow metal key", "white ceramic plate"], "negative_prompt": "low detail, color confusion, overlapping shadows", "diversity_tags": ["metal object", "ceramic plate"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "037b1ca0a450", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4286, "diversity_score": 0.3714, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_037b1ca0a450", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_037b1ca0a450/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28814754, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_037b1ca0a450/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_037b1ca0a450/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_037b1ca0a450/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9392488, "qwen_elapsed_seconds": 5.139, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_037b1ca0a450/record.json"} +{"prompt_id": "occlusion_overlap_d5_040632148059", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic setup of a blue glass vial partially obscured by a gray foam block, clean separation between layers", "target_evidence": "blue glass vial", "expected_visual_elements": ["blue glass vial", "gray foam block"], "negative_prompt": "low detail, overlapping textures, color blending", "diversity_tags": ["glass object", "foam occluder"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "040632148059", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.3667, "diversity_score": 0.4333, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_040632148059", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_040632148059/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29865194, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_040632148059/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_040632148059/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_040632148059/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1247382, "qwen_elapsed_seconds": 5.173, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_040632148059/record.json"} +{"prompt_id": "occlusion_overlap_d5_0b548f6daac9", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic scene of a green pen partially hidden behind a white ceramic mug on a wooden desk, high contrast edges", "target_evidence": "green pen", "expected_visual_elements": ["green pen", "white mug"], "negative_prompt": "low resolution, poor focus, messy arrangement, unclear depth", "diversity_tags": ["ceramic mug", "wooden desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0b548f6daac9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4483, "diversity_score": 0.3517, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_0b548f6daac9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_0b548f6daac9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22770948, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_0b548f6daac9/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_0b548f6daac9/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_0b548f6daac9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7257538, "qwen_elapsed_seconds": 5.229, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_0b548f6daac9/record.json"} +{"prompt_id": "occlusion_overlap_d5_220e55f3cadd", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic setup of a blue bottle partially covered by a green fabric cloth on a metal counter, sharp focus", "target_evidence": "blue bottle", "expected_visual_elements": ["blue bottle", "green fabric cloth"], "negative_prompt": "low detail, poor lighting, mixed textures, chaotic background", "diversity_tags": ["fabric cloth", "metal counter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "220e55f3cadd", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_220e55f3cadd", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_220e55f3cadd/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23147168, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_220e55f3cadd/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_220e55f3cadd/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_220e55f3cadd/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9620109, "qwen_elapsed_seconds": 5.248, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_220e55f3cadd/record.json"} +{"prompt_id": "occlusion_overlap_d5_34297a3f3945", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic scene of a white ceramic bowl with a black rubber band wrapped around it, clear foreground separation", "target_evidence": "white ceramic bowl", "expected_visual_elements": ["white ceramic bowl", "black rubber band"], "negative_prompt": "low detail, soft edges, color confusion", "diversity_tags": ["ceramic object", "rubber occluder"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "34297a3f3945", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_34297a3f3945", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_34297a3f3945/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28194364, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_34297a3f3945/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_34297a3f3945/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_34297a3f3945/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 856611, "qwen_elapsed_seconds": 5.136, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_34297a3f3945/record.json"} +{"prompt_id": "occlusion_overlap_d5_374850b6cc53", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic scene of a yellow key partially hidden behind a transparent glass plate on a wooden shelf, clear foreground separation", "target_evidence": "yellow key", "expected_visual_elements": ["yellow key", "transparent glass plate"], "negative_prompt": "low definition, poor focus, mixed colors, unclear edges", "diversity_tags": ["glass plate", "wooden shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "374850b6cc53", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_374850b6cc53", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_374850b6cc53/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21594890, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_374850b6cc53/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_374850b6cc53/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_374850b6cc53/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3195207, "qwen_elapsed_seconds": 5.169, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_374850b6cc53/record.json"} +{"prompt_id": "occlusion_overlap_d5_449af12f868d", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic setup of a brown book partially covered by a blue glass jar on a ceramic tile floor, clear foreground separation", "target_evidence": "brown book", "expected_visual_elements": ["brown book", "blue glass jar"], "negative_prompt": "low detail, unclear edges, mixed colors, poor lighting", "diversity_tags": ["glass jar", "ceramic tile"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "449af12f868d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4828, "diversity_score": 0.3172, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_449af12f868d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_449af12f868d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25238874, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_449af12f868d/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_449af12f868d/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_449af12f868d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 954668, "qwen_elapsed_seconds": 5.161, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_449af12f868d/record.json"} +{"prompt_id": "occlusion_overlap_d5_4950cac45f3a", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic setup of a silver watch partially hidden behind a stack of books on a stone counter, high contrast edges", "target_evidence": "silver watch", "expected_visual_elements": ["silver watch", "stack of books"], "negative_prompt": "low detail, unclear depth, messy arrangement, poor lighting", "diversity_tags": ["book stack", "stone counter"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4950cac45f3a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.4231, "diversity_score": 0.3769, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_4950cac45f3a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_4950cac45f3a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24706603, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_4950cac45f3a/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_4950cac45f3a/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_4950cac45f3a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9283257, "qwen_elapsed_seconds": 5.211, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_4950cac45f3a/record.json"} +{"prompt_id": "occlusion_overlap_d5_4e7fba724951", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a blue ceramic mug with a black rubber band wrapped around it, sharp focus on mug's handle", "target_evidence": "blue ceramic mug", "expected_visual_elements": ["blue ceramic mug", "black rubber band"], "negative_prompt": "low resolution, soft edges, color blending, shadows", "diversity_tags": ["opaque occlusion", "rubber band"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4e7fba724951", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_4e7fba724951", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_4e7fba724951/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26646983, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_4e7fba724951/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_4e7fba724951/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_4e7fba724951/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8970861, "qwen_elapsed_seconds": 5.189, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_4e7fba724951/record.json"} +{"prompt_id": "occlusion_overlap_d5_592e38383325", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a blue key lying under a transparent glass bowl on a metal shelf, clear foreground separation", "target_evidence": "blue key", "expected_visual_elements": ["blue key", "transparent glass bowl"], "negative_prompt": "low resolution, out-of-focus, shadow distortion, poor lighting", "diversity_tags": ["transparent overlap", "metal shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "592e38383325", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_592e38383325", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_592e38383325/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29022556, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_592e38383325/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_592e38383325/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_592e38383325/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3428479, "qwen_elapsed_seconds": 5.217, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_592e38383325/record.json"} +{"prompt_id": "occlusion_overlap_d5_750e938319b3", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic scene of a green wooden spoon partially covered by a black leather glove, clear foreground separation", "target_evidence": "green wooden spoon", "expected_visual_elements": ["green wooden spoon", "black leather glove"], "negative_prompt": "low contrast, soft edges, overlapping shadows", "diversity_tags": ["wooden object", "leather occluder"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "750e938319b3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4643, "diversity_score": 0.3357, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_750e938319b3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_750e938319b3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27954013, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_750e938319b3/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_750e938319b3/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_750e938319b3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5021956, "qwen_elapsed_seconds": 5.224, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_750e938319b3/record.json"} +{"prompt_id": "occlusion_overlap_d5_8e36697fdbcb", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a white plastic bottle with a brown paper towel draped over it, distinct layers with sharp focus", "target_evidence": "white plastic bottle", "expected_visual_elements": ["white plastic bottle", "brown paper towel"], "negative_prompt": "blurred edges, low resolution, color mixing", "diversity_tags": ["plastic object", "paper occluder"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "8e36697fdbcb", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.3793, "diversity_score": 0.4207, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_8e36697fdbcb", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_8e36697fdbcb/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27616808, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_8e36697fdbcb/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_8e36697fdbcb/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_8e36697fdbcb/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5219414, "qwen_elapsed_seconds": 5.216, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_8e36697fdbcb/record.json"} +{"prompt_id": "occlusion_overlap_d5_a1bd4de9f92a", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a black phone partially obscured by a red plastic box on a stone wall, clean boundaries", "target_evidence": "black phone", "expected_visual_elements": ["black phone", "red plastic box"], "negative_prompt": "low clarity, poor lighting, overlapping textures, unclear depth", "diversity_tags": ["plastic box", "stone wall"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a1bd4de9f92a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4643, "diversity_score": 0.3357, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_a1bd4de9f92a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_a1bd4de9f92a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27136583, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_a1bd4de9f92a/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_a1bd4de9f92a/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_a1bd4de9f92a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6990186, "qwen_elapsed_seconds": 5.224, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_a1bd4de9f92a/record.json"} +{"prompt_id": "occlusion_overlap_d5_ae173b4d208d", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a white notebook partially obscured by a green watering can on a green plant stand, sharp focus", "target_evidence": "white notebook", "expected_visual_elements": ["white notebook", "green watering can"], "negative_prompt": "low clarity, out-of-focus, poor lighting, chaotic background", "diversity_tags": ["plant stand", "watering can"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ae173b4d208d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4138, "diversity_score": 0.3862, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_ae173b4d208d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_ae173b4d208d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23061814, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_ae173b4d208d/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_ae173b4d208d/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_ae173b4d208d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1213427, "qwen_elapsed_seconds": 5.185, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_ae173b4d208d/record.json"} +{"prompt_id": "occlusion_overlap_d5_b965326767df", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic scene of a silver metal ruler with a black rubber eraser overlapping it, clear foreground and background", "target_evidence": "silver metal ruler", "expected_visual_elements": ["silver metal ruler", "black rubber eraser"], "negative_prompt": "blurry, color confusion, low contrast", "diversity_tags": ["metal object", "rubber overlap"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b965326767df", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.44, "diversity_score": 0.36, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_b965326767df", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_b965326767df/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22031841, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_b965326767df/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_b965326767df/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_b965326767df/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1019696, "qwen_elapsed_seconds": 5.219, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_b965326767df/record.json"} +{"prompt_id": "occlusion_overlap_d5_c88280e67603", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a brown wooden block with a green fabric cloth draped over it, clear boundary between objects", "target_evidence": "brown wooden block", "expected_visual_elements": ["brown wooden block", "green fabric cloth"], "negative_prompt": "low resolution, soft edges, color mixing", "diversity_tags": ["wooden object", "fabric occluder"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c88280e67603", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.3571, "diversity_score": 0.4429, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_c88280e67603", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_c88280e67603/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29554302, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_c88280e67603/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_c88280e67603/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_c88280e67603/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3885273, "qwen_elapsed_seconds": 5.194, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_c88280e67603/record.json"} +{"prompt_id": "occlusion_overlap_d5_dabe26d7ed60", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic scene of a red mug partially covered by a green leafy plant on a wooden table, clean boundaries, sharp focus", "target_evidence": "red mug", "expected_visual_elements": ["red mug", "green plant"], "negative_prompt": "blurry, low contrast, overlapping textures, chaotic background", "diversity_tags": ["color contrast", "plant occluder"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "dabe26d7ed60", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_dabe26d7ed60", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_dabe26d7ed60/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24993907, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_dabe26d7ed60/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_dabe26d7ed60/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_dabe26d7ed60/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6874906, "qwen_elapsed_seconds": 5.167, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_dabe26d7ed60/record.json"} +{"prompt_id": "occlusion_overlap_d5_f24bb6b46c68", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic scene of a yellow cup partially covered by a black fabric cloth on a white wooden floor, clean depth ordering", "target_evidence": "yellow cup", "expected_visual_elements": ["yellow cup", "black cloth"], "negative_prompt": "low definition, mixed textures, overlapping colors, poor focus", "diversity_tags": ["fabric occluder", "wooden floor"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f24bb6b46c68", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_f24bb6b46c68", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_f24bb6b46c68/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20868115, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_f24bb6b46c68/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_f24bb6b46c68/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_f24bb6b46c68/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9376378, "qwen_elapsed_seconds": 5.157, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_f24bb6b46c68/record.json"} +{"prompt_id": "occlusion_overlap_d5_f712406db181", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic scene of a red glass jar partially covered by a green fabric scarf, clear boundary between transparent and opaque layers", "target_evidence": "red glass jar", "expected_visual_elements": ["red glass jar", "green fabric scarf"], "negative_prompt": "blurry, low contrast, chaotic background, overlapping textures", "diversity_tags": ["transparent overlap", "fabric occluder"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f712406db181", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4, "diversity_score": 0.4, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_f712406db181", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_f712406db181/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20321539, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_f712406db181/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_f712406db181/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_f712406db181/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1877130, "qwen_elapsed_seconds": 5.202, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_f712406db181/record.json"} +{"prompt_id": "occlusion_overlap_d5_fde2f70f984b", "run_name": "flux2_qil_decomposition_pilot", "category": "occlusion_overlap", "difficulty": 5, "image_prompt": "synthetic photorealistic setup of a red plastic cup partially hidden under a black fabric cloth, sharp focus on cup's rim", "target_evidence": "red plastic cup", "expected_visual_elements": ["red plastic cup", "black fabric cloth"], "negative_prompt": "blurred edges, overlapping shadows, low contrast", "diversity_tags": ["plastic object", "fabric occluder"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "fde2f70f984b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic setup of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "occlusion_overlap_d5_fde2f70f984b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/occlusion_overlap/occlusion_overlap_d5_fde2f70f984b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26618263, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/occlusion_overlap/occlusion_overlap_d5_fde2f70f984b/layers/layer_00.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_fde2f70f984b/layers/layer_01.png", "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_fde2f70f984b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1045446, "qwen_elapsed_seconds": 5.245, "qwen_record_path": "layers-x/k3/occlusion_overlap/occlusion_overlap_d5_fde2f70f984b/record.json"} +{"prompt_id": "ocr_heavy_d1_000f11858c77", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic label on a bottle with 'KILO' in white on dark blue, high contrast and clean boundaries", "target_evidence": "text region with 'KILO'", "expected_visual_elements": ["bottle", "white text", "dark blue background"], "negative_prompt": "low resolution, blurry text, watermarks, real brands", "diversity_tags": ["label", "dark blue background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "000f11858c77", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic label on a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_000f11858c77", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_000f11858c77/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23437487, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_000f11858c77/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_000f11858c77/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_000f11858c77/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2280452, "qwen_elapsed_seconds": 5.201, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_000f11858c77/record.json"} +{"prompt_id": "ocr_heavy_d1_0c5876512174", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic jersey with 'ALTO' embroidered in white on a dark blue fabric, clean boundaries and sharp focus", "target_evidence": "short synthetic non-brand string 'ALTO'", "expected_visual_elements": ["jersey", "embroidered text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["jersey", "embroidered text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0c5876512174", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic jersey with alto", "max_similarity_to_other_prompt": 0.6522, "diversity_score": 0.1478, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_0c5876512174", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_0c5876512174/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25801013, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_0c5876512174/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_0c5876512174/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_0c5876512174/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9764836, "qwen_elapsed_seconds": 5.144, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_0c5876512174/record.json"} +{"prompt_id": "ocr_heavy_d1_13ab21d05c2c", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic jersey with DELTA, sharp focus on the front, clean boundaries, high contrast between text and fabric", "target_evidence": "text region with 'DELTA'", "expected_visual_elements": ["jersey", "text"], "negative_prompt": "low contrast, blurry text, real brands, watermarks, famous logos", "diversity_tags": ["jersey", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "13ab21d05c2c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic jersey with delta", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.48, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d1_13ab21d05c2c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_13ab21d05c2c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27217797, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_13ab21d05c2c/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_13ab21d05c2c/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_13ab21d05c2c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8487427, "qwen_elapsed_seconds": 5.253, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_13ab21d05c2c/record.json"} +{"prompt_id": "ocr_heavy_d1_2650afb8803e", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic menu board with 'EXIT B' written in white on a dark brown surface, clean boundaries and sharp focus", "target_evidence": "short synthetic non-brand string 'EXIT B'", "expected_visual_elements": ["menu board", "white text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["menu", "white text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2650afb8803e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic menu board with", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_2650afb8803e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_2650afb8803e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25805093, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_2650afb8803e/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_2650afb8803e/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_2650afb8803e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7562330, "qwen_elapsed_seconds": 5.232, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_2650afb8803e/record.json"} +{"prompt_id": "ocr_heavy_d1_33d8410adce1", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic jersey with 'ALTO' in white on a black sleeve, clearly visible and separated from surrounding fabric with high contrast and sharp focus", "target_evidence": "one target text region with a short synthetic non-brand string", "expected_visual_elements": ["jersey", "black sleeve", "white text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["jersey", "white text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "33d8410adce1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic jersey with alto", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_33d8410adce1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_33d8410adce1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25933927, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_33d8410adce1/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_33d8410adce1/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_33d8410adce1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3502692, "qwen_elapsed_seconds": 5.163, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_33d8410adce1/record.json"} +{"prompt_id": "ocr_heavy_d1_3856b00e6510", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic label on a metal can with 'BLUE-7' in black on a light blue background, clear foreground separation", "target_evidence": "short synthetic non-brand string 'BLUE-7'", "expected_visual_elements": ["label", "black text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["label", "black text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "3856b00e6510", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic label on a", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_3856b00e6510", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_3856b00e6510/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28296655, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_3856b00e6510/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_3856b00e6510/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_3856b00e6510/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8098161, "qwen_elapsed_seconds": 5.217, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_3856b00e6510/record.json"} +{"prompt_id": "ocr_heavy_d1_3f9acfe3afc8", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic crate with 'TAXI 5' printed in red on a clean wooden surface, high contrast for easy readability", "target_evidence": "short synthetic non-brand string 'TAXI 5'", "expected_visual_elements": ["crate", "red printed text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["crate", "red text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "3f9acfe3afc8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with taxi", "max_similarity_to_other_prompt": 0.56, "diversity_score": 0.24, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_3f9acfe3afc8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_3f9acfe3afc8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22030391, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_3f9acfe3afc8/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_3f9acfe3afc8/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_3f9acfe3afc8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8826057, "qwen_elapsed_seconds": 5.185, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_3f9acfe3afc8/record.json"} +{"prompt_id": "ocr_heavy_d1_56dbca040b6b", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic poster with 'ROOM 18' in red on a white background, sharp focus on the text area", "target_evidence": "short synthetic non-brand string 'ROOM 18'", "expected_visual_elements": ["poster", "red text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["poster", "red text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "56dbca040b6b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic poster with room", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.4348, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d1_56dbca040b6b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_56dbca040b6b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29728141, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_56dbca040b6b/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_56dbca040b6b/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_56dbca040b6b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1771051, "qwen_elapsed_seconds": 5.291, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_56dbca040b6b/record.json"} +{"prompt_id": "ocr_heavy_d1_6ebdfe2bb2f9", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic notebook page with clear 'UNIT 12' written in bold black font on a clean white background, sharp focus on text area", "target_evidence": "short synthetic non-brand string 'UNIT 12'", "expected_visual_elements": ["notebook page", "bold black text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["notebook", "bold text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6ebdfe2bb2f9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.7917, "diversity_score": 0.0083, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_6ebdfe2bb2f9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_6ebdfe2bb2f9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 30060673, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_6ebdfe2bb2f9/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_6ebdfe2bb2f9/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_6ebdfe2bb2f9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4191631, "qwen_elapsed_seconds": 5.226, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_6ebdfe2bb2f9/record.json"} +{"prompt_id": "ocr_heavy_d1_738ec3c08a8e", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic label on a shipping crate with clear high-contrast text 'UNIT 12', clean boundaries, sharp focus", "target_evidence": "text region with 'UNIT 12'", "expected_visual_elements": ["shipping crate", "label", "high-contrast text"], "negative_prompt": "low resolution, blurry text, watermarks, real brands", "diversity_tags": ["label", "crate", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "738ec3c08a8e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic label on a", "max_similarity_to_other_prompt": 0.6522, "diversity_score": 0.1478, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_738ec3c08a8e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_738ec3c08a8e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21133861, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_738ec3c08a8e/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_738ec3c08a8e/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_738ec3c08a8e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9321658, "qwen_elapsed_seconds": 5.218, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_738ec3c08a8e/record.json"} +{"prompt_id": "ocr_heavy_d1_7f213a71014d", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic tag on a package with 'GATE 42' in white text on black background, clean edges and high contrast", "target_evidence": "text region with 'GATE 42'", "expected_visual_elements": ["package", "white text", "black background"], "negative_prompt": "low resolution, blurry text, watermarks, real brands", "diversity_tags": ["tag", "black background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7f213a71014d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tag on a", "max_similarity_to_other_prompt": 0.6957, "diversity_score": 0.1043, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_7f213a71014d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_7f213a71014d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28552416, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_7f213a71014d/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_7f213a71014d/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_7f213a71014d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2550107, "qwen_elapsed_seconds": 5.3, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_7f213a71014d/record.json"} +{"prompt_id": "ocr_heavy_d1_a25d6db04040", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic crate with 'ZEN' printed in red on a white label, positioned at the top center with clean boundaries and sharp focus for easy text recognition", "target_evidence": "one target text region with a short synthetic non-brand string", "expected_visual_elements": ["crate", "white label", "red text"], "negative_prompt": "faded text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["crate", "red text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a25d6db04040", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with zen", "max_similarity_to_other_prompt": 0.6071, "diversity_score": 0.3929, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d1_a25d6db04040", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_a25d6db04040/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21502967, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_a25d6db04040/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_a25d6db04040/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_a25d6db04040/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 358917, "qwen_elapsed_seconds": 5.274, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_a25d6db04040/record.json"} +{"prompt_id": "ocr_heavy_d1_b0907cc9c496", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic screen displaying 'UNIT 12' with high contrast white text on dark blue background, clean boundaries, sharp focus", "target_evidence": "UNIT 12", "expected_visual_elements": ["screen", "text region"], "negative_prompt": "low resolution, blurred text, real brand logos, watermarks, distorted edges", "diversity_tags": ["screen", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b0907cc9c496", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying unit", "max_similarity_to_other_prompt": 0.85, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_b0907cc9c496", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_b0907cc9c496/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21766730, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_b0907cc9c496/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_b0907cc9c496/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_b0907cc9c496/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5637403, "qwen_elapsed_seconds": 5.246, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_b0907cc9c496/record.json"} +{"prompt_id": "ocr_heavy_d1_b0b7cb1f692c", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic screen displaying UNIT 12 with clean boundaries and high contrast, sharp focus on the text, no background distractions", "target_evidence": "text region with 'UNIT 12'", "expected_visual_elements": ["screen", "text"], "negative_prompt": "low contrast, blurry text, real brands, watermarks, famous logos", "diversity_tags": ["screen", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b0b7cb1f692c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying unit", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_b0b7cb1f692c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_b0b7cb1f692c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22268026, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_b0b7cb1f692c/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_b0b7cb1f692c/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_b0b7cb1f692c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6236659, "qwen_elapsed_seconds": 5.413, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_b0b7cb1f692c/record.json"} +{"prompt_id": "ocr_heavy_d1_b2d96a09574c", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic sign with 'ROOM 18' in large white letters on black background, clear foreground separation, high contrast", "target_evidence": "ROOM 18", "expected_visual_elements": ["sign", "text region"], "negative_prompt": "low resolution, blurred text, real brand logos, watermarks, distorted edges", "diversity_tags": ["sign", "large text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b2d96a09574c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic sign with room", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_b2d96a09574c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_b2d96a09574c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 30207610, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_b2d96a09574c/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_b2d96a09574c/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_b2d96a09574c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4932930, "qwen_elapsed_seconds": 5.316, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_b2d96a09574c/record.json"} +{"prompt_id": "ocr_heavy_d1_bc6a582b1c0a", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic screen displaying 'DELTA' in bold black letters against a dark blue background with clean edges and clear separation from surrounding elements", "target_evidence": "one target text region with a short synthetic non-brand string", "expected_visual_elements": ["screen", "dark blue background", "bold black letters"], "negative_prompt": "distorted text, low resolution, watermarks, real brands, famous logos", "diversity_tags": ["screen", "bold letters"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "bc6a582b1c0a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying delta", "max_similarity_to_other_prompt": 0.5185, "diversity_score": 0.2815, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_bc6a582b1c0a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_bc6a582b1c0a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25774410, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_bc6a582b1c0a/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_bc6a582b1c0a/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_bc6a582b1c0a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4719011, "qwen_elapsed_seconds": 6.734, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_bc6a582b1c0a/record.json"} +{"prompt_id": "ocr_heavy_d1_c4c8a35de64c", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic screen displaying 'UNIT 12' with high contrast text on a dark background, clean boundaries, sharp focus", "target_evidence": "UNIT 12", "expected_visual_elements": ["screen", "dark background"], "negative_prompt": "blurry text, low contrast, real brands, watermarks", "diversity_tags": ["screen", "dark background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c4c8a35de64c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying unit", "max_similarity_to_other_prompt": 1.0, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_c4c8a35de64c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_c4c8a35de64c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20888916, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_c4c8a35de64c/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_c4c8a35de64c/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_c4c8a35de64c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2355848, "qwen_elapsed_seconds": 6.318, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_c4c8a35de64c/record.json"} +{"prompt_id": "ocr_heavy_d1_cd65b7da73f0", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic sign with 'ROOM 18' in red letters on a white board, clear boundaries and high contrast", "target_evidence": "text region with 'ROOM 18'", "expected_visual_elements": ["sign", "red letters", "white board"], "negative_prompt": "low resolution, blurry text, watermarks, real brands", "diversity_tags": ["sign", "red letters"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cd65b7da73f0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic sign with room", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_cd65b7da73f0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_cd65b7da73f0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21981990, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_cd65b7da73f0/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_cd65b7da73f0/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_cd65b7da73f0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8380900, "qwen_elapsed_seconds": 5.717, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_cd65b7da73f0/record.json"} +{"prompt_id": "ocr_heavy_d1_d55b17070937", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic notebook page with clear front-facing text 'UNIT 12' on a clean white background, sharp focus and high contrast for easy OCR extraction", "target_evidence": "one target text region with a short synthetic non-brand string", "expected_visual_elements": ["notebook page", "white background", "sharp focus"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["notebook", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d55b17070937", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.8462, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_d55b17070937", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_d55b17070937/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21303860, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_d55b17070937/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_d55b17070937/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_d55b17070937/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9572157, "qwen_elapsed_seconds": 5.787, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_d55b17070937/record.json"} +{"prompt_id": "ocr_heavy_d1_f7f1661a5836", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 1, "image_prompt": "synthetic photorealistic tag on a wooden shelf with 'GATE 42' in black on a light yellow background, clean boundaries", "target_evidence": "short synthetic non-brand string 'GATE 42'", "expected_visual_elements": ["tag", "black text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["tag", "black text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f7f1661a5836", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tag on a", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d1_f7f1661a5836", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d1_f7f1661a5836/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21890553, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d1_f7f1661a5836/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_f7f1661a5836/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d1_f7f1661a5836/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6782305, "qwen_elapsed_seconds": 6.518, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d1_f7f1661a5836/record.json"} +{"prompt_id": "ocr_heavy_d2_0a4780854e35", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic poster with green text 'MESA' on a white wall, sharp focus and clear separation for OCR testing", "target_evidence": "text region with 'MESA'", "expected_visual_elements": ["poster", "white wall", "green text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["poster", "text color", "wall type"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0a4780854e35", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic poster with green", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.4615, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d2_0a4780854e35", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_0a4780854e35/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25368714, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_0a4780854e35/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_0a4780854e35/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_0a4780854e35/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8084466, "qwen_elapsed_seconds": 5.478, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_0a4780854e35/record.json"} +{"prompt_id": "ocr_heavy_d2_139b194f7e8d", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic jersey with black text 'TAXI 5' on a white background, clean boundaries, sharp focus on the text area", "target_evidence": "text region with 'TAXI 5'", "expected_visual_elements": ["jersey", "white background", "black text"], "negative_prompt": "faded text, background distractions, low contrast", "diversity_tags": ["jersey", "text on fabric", "clean boundaries"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "139b194f7e8d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic jersey with black", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.375, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d2_139b194f7e8d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_139b194f7e8d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29697394, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_139b194f7e8d/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_139b194f7e8d/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_139b194f7e8d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3232545, "qwen_elapsed_seconds": 5.908, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_139b194f7e8d/record.json"} +{"prompt_id": "ocr_heavy_d2_1956900f9016", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic package with white text 'ROOM 18' on a green label, high contrast and clean edges for OCR testing", "target_evidence": "text region with 'ROOM 18'", "expected_visual_elements": ["package", "green label", "white text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["package", "label", "text color"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1956900f9016", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic package with white", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_1956900f9016", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_1956900f9016/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25834324, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_1956900f9016/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_1956900f9016/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_1956900f9016/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2148945, "qwen_elapsed_seconds": 5.934, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_1956900f9016/record.json"} +{"prompt_id": "ocr_heavy_d2_217c0e605027", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic screen displaying 'UNIT 12' with high contrast text on a control panel, clean boundaries, sharp focus", "target_evidence": "text region with 'UNIT 12'", "expected_visual_elements": ["control panel", "high contrast text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands", "diversity_tags": ["screen", "control panel", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "217c0e605027", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying unit", "max_similarity_to_other_prompt": 0.8, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_217c0e605027", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_217c0e605027/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23845593, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_217c0e605027/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_217c0e605027/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_217c0e605027/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1638444, "qwen_elapsed_seconds": 5.838, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_217c0e605027/record.json"} +{"prompt_id": "ocr_heavy_d2_240ab77bdd32", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic screen displaying 'EXIT B' on a dark blue interface with clean borders and high contrast for OCR", "target_evidence": "text region with 'EXIT B'", "expected_visual_elements": ["screen", "dark blue interface", "white text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["screen", "interface", "text color"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "240ab77bdd32", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying exit", "max_similarity_to_other_prompt": 0.6818, "diversity_score": 0.3182, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d2_240ab77bdd32", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_240ab77bdd32/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28928117, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_240ab77bdd32/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_240ab77bdd32/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_240ab77bdd32/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7598655, "qwen_elapsed_seconds": 5.588, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_240ab77bdd32/record.json"} +{"prompt_id": "ocr_heavy_d2_31a1daae64b9", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic package with white text 'ALTO' on a dark green background, high contrast, sharp focus on the text area", "target_evidence": "text region with 'ALTO'", "expected_visual_elements": ["package", "dark green background", "white text"], "negative_prompt": "faded text, low contrast, background distractions", "diversity_tags": ["package", "text on surface", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "31a1daae64b9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic package with white", "max_similarity_to_other_prompt": 0.5217, "diversity_score": 0.2783, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_31a1daae64b9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_31a1daae64b9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20985338, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_31a1daae64b9/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_31a1daae64b9/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_31a1daae64b9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2057430, "qwen_elapsed_seconds": 5.741, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_31a1daae64b9/record.json"} +{"prompt_id": "ocr_heavy_d2_428995c0f910", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic tag on a metal object with red text 'ZEN' on a clean white surface, sharp focus and clear separation for OCR", "target_evidence": "text region with 'ZEN'", "expected_visual_elements": ["tag", "metal object", "red text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["tag", "object type", "text color"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "428995c0f910", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tag on a", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_428995c0f910", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_428995c0f910/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28762524, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_428995c0f910/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_428995c0f910/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_428995c0f910/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3104591, "qwen_elapsed_seconds": 5.388, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_428995c0f910/record.json"} +{"prompt_id": "ocr_heavy_d2_4c156ae47ca2", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic notebook page with clear text 'UNIT 12' on a clean white background, sharp focus, high contrast between dark text and light paper, no distractions", "target_evidence": "short synthetic non-brand string 'UNIT 12'", "expected_visual_elements": ["notebook page", "white background", "dark text", "clean boundaries"], "negative_prompt": "low contrast, blurred text, real brand logos, watermarks, complex background patterns", "diversity_tags": ["notebook", "text on paper", "high contrast", "clean boundaries"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4c156ae47ca2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.7308, "diversity_score": 0.0692, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_4c156ae47ca2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_4c156ae47ca2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24219798, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_4c156ae47ca2/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_4c156ae47ca2/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_4c156ae47ca2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2708378, "qwen_elapsed_seconds": 5.925, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_4c156ae47ca2/record.json"} +{"prompt_id": "ocr_heavy_d2_5304c1b910b9", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic crate with bright yellow label 'DELTA' on a red background, clear boundaries, sharp focus on the label area", "target_evidence": "text region with 'DELTA'", "expected_visual_elements": ["crate", "red background", "yellow label"], "negative_prompt": "distorted text, overlapping elements, low resolution", "diversity_tags": ["crate", "label on surface", "color contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5304c1b910b9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with bright", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_5304c1b910b9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_5304c1b910b9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25242857, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_5304c1b910b9/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_5304c1b910b9/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_5304c1b910b9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7360831, "qwen_elapsed_seconds": 5.648, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_5304c1b910b9/record.json"} +{"prompt_id": "ocr_heavy_d2_73ba33364e01", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic sign with yellow text 'GATE 42' on a black metal frame, high contrast and clean boundaries for OCR", "target_evidence": "text region with 'GATE 42'", "expected_visual_elements": ["sign", "black metal frame", "yellow text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["sign", "frame type", "text color"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "73ba33364e01", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic sign with yellow", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_73ba33364e01", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_73ba33364e01/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29558039, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_73ba33364e01/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_73ba33364e01/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_73ba33364e01/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1375565, "qwen_elapsed_seconds": 5.841, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_73ba33364e01/record.json"} +{"prompt_id": "ocr_heavy_d2_947d1f772a51", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic notebook page with clean white text 'UNIT 12' on top right, sharp focus, high contrast against dark paper", "target_evidence": "short synthetic non-brand string 'UNIT 12'", "expected_visual_elements": ["notebook page", "white text on dark paper"], "negative_prompt": "blurry text, low contrast, real brand logos, watermarks", "diversity_tags": ["notebook", "text placement", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "947d1f772a51", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.72, "diversity_score": 0.08, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_947d1f772a51", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_947d1f772a51/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24642220, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_947d1f772a51/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_947d1f772a51/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_947d1f772a51/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7613937, "qwen_elapsed_seconds": 5.847, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_947d1f772a51/record.json"} +{"prompt_id": "ocr_heavy_d2_96360ca89560", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic sign with 'EXIT B' on a green background, clear foreground separation, sharp focus on the text area", "target_evidence": "text region with 'EXIT B'", "expected_visual_elements": ["sign", "green background", "white text"], "negative_prompt": "overlapping text, low contrast, distorted edges", "diversity_tags": ["sign", "text on board", "foreground separation"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "96360ca89560", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic sign with exit", "max_similarity_to_other_prompt": 0.5217, "diversity_score": 0.2783, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_96360ca89560", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_96360ca89560/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28064991, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_96360ca89560/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_96360ca89560/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_96360ca89560/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8696512, "qwen_elapsed_seconds": 5.546, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_96360ca89560/record.json"} +{"prompt_id": "ocr_heavy_d2_bbbf13acc736", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic notebook page with clear blue text 'UNIT 12' on a clean white background, sharp focus and high contrast for easy OCR", "target_evidence": "text region with 'UNIT 12'", "expected_visual_elements": ["notebook page", "white background", "blue text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["notebook", "text color", "background type"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "bbbf13acc736", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.8462, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_bbbf13acc736", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_bbbf13acc736/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27850526, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_bbbf13acc736/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_bbbf13acc736/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_bbbf13acc736/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2758237, "qwen_elapsed_seconds": 5.364, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_bbbf13acc736/record.json"} +{"prompt_id": "ocr_heavy_d2_c2f79ba07cc4", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic screen displaying 'ORBIT' on a light gray interface with clean borders and high contrast for OCR", "target_evidence": "text region with 'ORBIT'", "expected_visual_elements": ["screen", "light gray interface", "black text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["screen", "interface", "text color"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c2f79ba07cc4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying orbit", "max_similarity_to_other_prompt": 0.6818, "diversity_score": 0.3182, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d2_c2f79ba07cc4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_c2f79ba07cc4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27629166, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_c2f79ba07cc4/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_c2f79ba07cc4/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_c2f79ba07cc4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 166372, "qwen_elapsed_seconds": 5.327, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_c2f79ba07cc4/record.json"} +{"prompt_id": "ocr_heavy_d2_d13eb1a9b43a", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic notebook page with clean white text 'UNIT 12' on a dark blue cover, sharp focus on the text area, high contrast for readability", "target_evidence": "text region with 'UNIT 12'", "expected_visual_elements": ["notebook page", "dark blue cover", "white text"], "negative_prompt": "blurry text, low contrast, watermarks, real brand logos", "diversity_tags": ["notebook", "text on cover", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d13eb1a9b43a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.6429, "diversity_score": 0.1571, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_d13eb1a9b43a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_d13eb1a9b43a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29541836, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_d13eb1a9b43a/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_d13eb1a9b43a/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_d13eb1a9b43a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3269319, "qwen_elapsed_seconds": 6.234, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_d13eb1a9b43a/record.json"} +{"prompt_id": "ocr_heavy_d2_de0b6525adc8", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic label on a bottle with 'LIMA' in white on black, sharp focus, clear foreground, clean lines", "target_evidence": "text region with 'LIMA'", "expected_visual_elements": ["bottle", "label"], "negative_prompt": "blurry text, low contrast, watermarks, logos", "diversity_tags": ["bottle", "label", "clean lines"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "de0b6525adc8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic label on a", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_de0b6525adc8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_de0b6525adc8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27522106, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_de0b6525adc8/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_de0b6525adc8/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_de0b6525adc8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8310427, "qwen_elapsed_seconds": 5.375, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_de0b6525adc8/record.json"} +{"prompt_id": "ocr_heavy_d2_de8099f74ce2", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic menu board with black text 'TAXI 5' on a light gray background, high contrast and clean edges for OCR", "target_evidence": "text region with 'TAXI 5'", "expected_visual_elements": ["menu board", "light gray background", "black text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["menu", "background color", "text color"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "de8099f74ce2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic menu board with", "max_similarity_to_other_prompt": 0.5185, "diversity_score": 0.2815, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_de8099f74ce2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_de8099f74ce2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27904448, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_de8099f74ce2/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_de8099f74ce2/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_de8099f74ce2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4206441, "qwen_elapsed_seconds": 5.362, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_de8099f74ce2/record.json"} +{"prompt_id": "ocr_heavy_d2_e6483fbe28ae", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic crate with red sticker 'DELTA' on a clean wooden surface, high contrast and clear boundaries for OCR testing", "target_evidence": "text region with 'DELTA'", "expected_visual_elements": ["crate", "wooden surface", "red sticker"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["crate", "sticker", "color contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e6483fbe28ae", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with red", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.48, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d2_e6483fbe28ae", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_e6483fbe28ae/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21267583, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_e6483fbe28ae/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_e6483fbe28ae/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_e6483fbe28ae/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5607310, "qwen_elapsed_seconds": 5.625, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_e6483fbe28ae/record.json"} +{"prompt_id": "ocr_heavy_d2_eaf3c28e9359", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic jersey with white text 'ALTO' on a dark blue fabric, sharp focus and clean edges for clear OCR", "target_evidence": "text region with 'ALTO'", "expected_visual_elements": ["jersey", "dark blue fabric", "white text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["jersey", "text color", "fabric type"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "eaf3c28e9359", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic jersey with white", "max_similarity_to_other_prompt": 0.6522, "diversity_score": 0.3478, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d2_eaf3c28e9359", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_eaf3c28e9359/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25150267, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_eaf3c28e9359/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_eaf3c28e9359/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_eaf3c28e9359/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9232055, "qwen_elapsed_seconds": 6.321, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_eaf3c28e9359/record.json"} +{"prompt_id": "ocr_heavy_d2_f46974a5b264", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 2, "image_prompt": "synthetic photorealistic menu board with 'ROOM 18' on a light brown background, sharp focus on the text area, clear boundaries", "target_evidence": "text region with 'ROOM 18'", "expected_visual_elements": ["menu board", "light brown background", "black text"], "negative_prompt": "overlapping elements, low contrast, watermarks", "diversity_tags": ["menu board", "text on surface", "clear boundaries"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f46974a5b264", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic menu board with", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d2_f46974a5b264", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d2_f46974a5b264/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29630976, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d2_f46974a5b264/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_f46974a5b264/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d2_f46974a5b264/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6269449, "qwen_elapsed_seconds": 5.33, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d2_f46974a5b264/record.json"} +{"prompt_id": "ocr_heavy_d3_26ba781eaff6", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "Synthetic photorealistic sign with 'EXIT B' in large yellow letters on a dark red background, high contrast, and sharp focus.", "target_evidence": "EXIT B", "expected_visual_elements": ["sign", "yellow text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["sign", "yellow text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "26ba781eaff6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic sign with exit", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_26ba781eaff6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_26ba781eaff6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24175876, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_26ba781eaff6/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_26ba781eaff6/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_26ba781eaff6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8967874, "qwen_elapsed_seconds": 5.305, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_26ba781eaff6/record.json"} +{"prompt_id": "ocr_heavy_d3_2b69cc14ecab", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "synthetic photorealistic crate with 'RIVER 9' printed in black on a light green surface, clean boundaries, and sharp focus for clear text", "target_evidence": "text region with 'RIVER 9'", "expected_visual_elements": ["crate", "light green surface", "black printed text"], "negative_prompt": "faded text, low contrast, complex background, watermarks, real brands", "diversity_tags": ["crate", "light green surface", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2b69cc14ecab", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with river", "max_similarity_to_other_prompt": 0.5556, "diversity_score": 0.2444, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_2b69cc14ecab", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_2b69cc14ecab/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26571522, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_2b69cc14ecab/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_2b69cc14ecab/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_2b69cc14ecab/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2966597, "qwen_elapsed_seconds": 5.343, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_2b69cc14ecab/record.json"} +{"prompt_id": "ocr_heavy_d3_4aecc9d7d690", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "synthetic photorealistic notebook page with clean text 'UNIT 12' on a white label, sharp focus, high contrast, clear foreground separation", "target_evidence": "text region with 'UNIT 12'", "expected_visual_elements": ["notebook page", "white label"], "negative_prompt": "blurry text, low contrast, overlapping elements, real brands, watermarks", "diversity_tags": ["notebook", "label", "white", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4aecc9d7d690", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.7391, "diversity_score": 0.0609, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_4aecc9d7d690", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_4aecc9d7d690/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27972653, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_4aecc9d7d690/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_4aecc9d7d690/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_4aecc9d7d690/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 898076, "qwen_elapsed_seconds": 5.53, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_4aecc9d7d690/record.json"} +{"prompt_id": "ocr_heavy_d3_4f8dff5eea21", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "synthetic photorealistic tag on a crate with 'ALTO' in bold letters, sharp focus, high contrast, and clean boundaries", "target_evidence": "text region with 'ALTO'", "expected_visual_elements": ["tag", "crate", "bold letters"], "negative_prompt": "faded text, low resolution, watermarks, logos, real brands", "diversity_tags": ["tag", "crate", "bold letters"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4f8dff5eea21", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tag on a", "max_similarity_to_other_prompt": 0.5833, "diversity_score": 0.2167, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_4f8dff5eea21", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_4f8dff5eea21/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29355218, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_4f8dff5eea21/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_4f8dff5eea21/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_4f8dff5eea21/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2368001, "qwen_elapsed_seconds": 5.91, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_4f8dff5eea21/record.json"} +{"prompt_id": "ocr_heavy_d3_541bd7c8bbf4", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "synthetic photorealistic sign with 'ZEN' in a clean white font, sharp focus, and high contrast against dark metal background", "target_evidence": "text region with 'ZEN'", "expected_visual_elements": ["sign", "white font", "dark metal"], "negative_prompt": "faded text, low resolution, watermarks, logos, real brands", "diversity_tags": ["sign", "white font", "dark metal"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "541bd7c8bbf4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic sign with zen", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.375, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d3_541bd7c8bbf4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_541bd7c8bbf4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23253157, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_541bd7c8bbf4/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_541bd7c8bbf4/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_541bd7c8bbf4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3495915, "qwen_elapsed_seconds": 5.612, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_541bd7c8bbf4/record.json"} +{"prompt_id": "ocr_heavy_d3_56d524c11b80", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "synthetic photorealistic notebook page with clear text 'UNIT 12' on a white label, clean boundaries, sharp focus, and high contrast against dark paper", "target_evidence": "text region with 'UNIT 12'", "expected_visual_elements": ["notebook page", "white label", "dark paper"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, logos", "diversity_tags": ["notebook", "label", "dark paper"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "56d524c11b80", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.7917, "diversity_score": 0.0083, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_56d524c11b80", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_56d524c11b80/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27399142, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_56d524c11b80/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_56d524c11b80/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_56d524c11b80/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9871924, "qwen_elapsed_seconds": 5.228, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_56d524c11b80/record.json"} +{"prompt_id": "ocr_heavy_d3_7ff15a31bf5c", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "synthetic photorealistic package with 'ORBIT' on a yellow label, clear boundaries, sharp focus, and high contrast against black background", "target_evidence": "text region with 'ORBIT'", "expected_visual_elements": ["package", "yellow label", "black background"], "negative_prompt": "blurry text, low contrast, watermarks, logos, real brands", "diversity_tags": ["package", "label", "black background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7ff15a31bf5c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic package with orbit", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_7ff15a31bf5c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_7ff15a31bf5c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21192273, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_7ff15a31bf5c/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_7ff15a31bf5c/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_7ff15a31bf5c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9783653, "qwen_elapsed_seconds": 5.739, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_7ff15a31bf5c/record.json"} +{"prompt_id": "ocr_heavy_d3_8d2d00cc5ae1", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "Synthetic photorealistic package with 'BLUE-7' in white on a dark orange surface, high contrast, and clean boundaries.", "target_evidence": "BLUE-7", "expected_visual_elements": ["package", "white text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["package", "white text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "8d2d00cc5ae1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic package with blue", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.3333, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d3_8d2d00cc5ae1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_8d2d00cc5ae1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22957981, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_8d2d00cc5ae1/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_8d2d00cc5ae1/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_8d2d00cc5ae1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4405772, "qwen_elapsed_seconds": 6.087, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_8d2d00cc5ae1/record.json"} +{"prompt_id": "ocr_heavy_d3_94f72b4b17f4", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "Synthetic photorealistic menu board with 'ROOM 18' in bright green on a wooden surface, clean boundaries, and high contrast.", "target_evidence": "ROOM 18", "expected_visual_elements": ["menu board", "green text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["menu board", "green text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "94f72b4b17f4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic menu board with", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_94f72b4b17f4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_94f72b4b17f4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25590825, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_94f72b4b17f4/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_94f72b4b17f4/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_94f72b4b17f4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7710868, "qwen_elapsed_seconds": 5.317, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_94f72b4b17f4/record.json"} +{"prompt_id": "ocr_heavy_d3_9d40a4593edf", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "synthetic photorealistic poster with 'DELTA' in bold red letters on a dark blue wall, clean edges, and high contrast for clear readability", "target_evidence": "text region with 'DELTA'", "expected_visual_elements": ["poster", "dark blue wall", "bold red letters"], "negative_prompt": "faded text, low contrast, complex background, watermarks, real brands", "diversity_tags": ["poster", "dark blue wall", "bold red letters"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "9d40a4593edf", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic poster with delta", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_9d40a4593edf", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_9d40a4593edf/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23519149, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_9d40a4593edf/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_9d40a4593edf/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_9d40a4593edf/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1491798, "qwen_elapsed_seconds": 5.92, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_9d40a4593edf/record.json"} +{"prompt_id": "ocr_heavy_d3_a05ae6b7ae6d", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "Synthetic photorealistic screen displaying 'KILO' in blue on a dark gray background, sharp focus, and clear foreground separation.", "target_evidence": "KILO", "expected_visual_elements": ["screen", "blue text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["screen", "blue text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a05ae6b7ae6d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying kilo", "max_similarity_to_other_prompt": 0.5833, "diversity_score": 0.4167, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d3_a05ae6b7ae6d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_a05ae6b7ae6d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27638904, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_a05ae6b7ae6d/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_a05ae6b7ae6d/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_a05ae6b7ae6d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1715843, "qwen_elapsed_seconds": 5.886, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_a05ae6b7ae6d/record.json"} +{"prompt_id": "ocr_heavy_d3_b7bb97875025", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "Synthetic photorealistic crate with 'TAXI 5' in red on a dark green surface, clean boundaries, high contrast, and sharp focus.", "target_evidence": "TAXI 5", "expected_visual_elements": ["crate", "red text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["crate", "red text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b7bb97875025", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with taxi", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_b7bb97875025", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_b7bb97875025/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29535886, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_b7bb97875025/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_b7bb97875025/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_b7bb97875025/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5805169, "qwen_elapsed_seconds": 5.363, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_b7bb97875025/record.json"} +{"prompt_id": "ocr_heavy_d3_bfbcb067ec4d", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "synthetic photorealistic notebook page with clear text 'UNIT 12' on a clean white background, sharp focus, high contrast, and clean boundaries", "target_evidence": "text region with 'UNIT 12'", "expected_visual_elements": ["notebook page", "white background", "clean text"], "negative_prompt": "blurry text, low contrast, complex background, watermarks, real brands", "diversity_tags": ["notebook", "white background", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "bfbcb067ec4d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.7917, "diversity_score": 0.0083, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_bfbcb067ec4d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_bfbcb067ec4d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22276586, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_bfbcb067ec4d/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_bfbcb067ec4d/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_bfbcb067ec4d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 654810, "qwen_elapsed_seconds": 5.482, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_bfbcb067ec4d/record.json"} +{"prompt_id": "ocr_heavy_d3_c1df9d4f9f04", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "synthetic photorealistic jersey with 'ALTO' in white letters on a black sleeve, high contrast, and clear foreground separation for readability", "target_evidence": "text region with 'ALTO'", "expected_visual_elements": ["jersey", "black sleeve", "white text"], "negative_prompt": "faded text, low contrast, complex background, watermarks, real brands", "diversity_tags": ["jersey", "black sleeve", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c1df9d4f9f04", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic jersey with alto", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_c1df9d4f9f04", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_c1df9d4f9f04/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24440224, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_c1df9d4f9f04/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_c1df9d4f9f04/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_c1df9d4f9f04/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7901654, "qwen_elapsed_seconds": 5.692, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_c1df9d4f9f04/record.json"} +{"prompt_id": "ocr_heavy_d3_c605a096c1b8", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "synthetic photorealistic crate with 'TAXI 5' on a bright red sticker, clean boundaries, sharp focus, and high contrast", "target_evidence": "text region with 'TAXI 5'", "expected_visual_elements": ["crate", "bright red sticker", "high contrast"], "negative_prompt": "faded text, low resolution, watermarks, logos, real brands", "diversity_tags": ["crate", "sticker", "bright red"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c605a096c1b8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with taxi", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_c605a096c1b8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_c605a096c1b8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27520937, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_c605a096c1b8/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_c605a096c1b8/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_c605a096c1b8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 220953, "qwen_elapsed_seconds": 5.57, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_c605a096c1b8/record.json"} +{"prompt_id": "ocr_heavy_d3_cb507dbacbdc", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "Synthetic photorealistic notebook page with 'UNIT 12' in bold black text on a clean white background, sharp focus, clear foreground separation.", "target_evidence": "UNIT 12", "expected_visual_elements": ["notebook page", "bold black text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["notebook", "black text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cb507dbacbdc", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.7917, "diversity_score": 0.0083, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_cb507dbacbdc", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_cb507dbacbdc/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22835604, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_cb507dbacbdc/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_cb507dbacbdc/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_cb507dbacbdc/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3278500, "qwen_elapsed_seconds": 5.66, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_cb507dbacbdc/record.json"} +{"prompt_id": "ocr_heavy_d3_d6ed8e69262a", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "Synthetic photorealistic tag on a wooden crate with 'ALTO' in black on a light brown background, clean boundaries, and clear foreground separation.", "target_evidence": "ALTO", "expected_visual_elements": ["tag", "black text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["tag", "black text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d6ed8e69262a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tag on a", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_d6ed8e69262a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_d6ed8e69262a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29139701, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_d6ed8e69262a/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_d6ed8e69262a/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_d6ed8e69262a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2488406, "qwen_elapsed_seconds": 5.805, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_d6ed8e69262a/record.json"} +{"prompt_id": "ocr_heavy_d3_e5718cdad61d", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "Synthetic photorealistic jersey with 'ORBIT' in white on a dark navy background, high contrast, clear foreground separation, and sharp focus.", "target_evidence": "ORBIT", "expected_visual_elements": ["jersey", "white text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["jersey", "white text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e5718cdad61d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic jersey with orbit", "max_similarity_to_other_prompt": 0.6522, "diversity_score": 0.3478, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d3_e5718cdad61d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_e5718cdad61d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 30142620, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_e5718cdad61d/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_e5718cdad61d/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_e5718cdad61d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5371076, "qwen_elapsed_seconds": 5.771, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_e5718cdad61d/record.json"} +{"prompt_id": "ocr_heavy_d3_ed267d7e662f", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "synthetic photorealistic menu board with 'ROOM 18' in a clean white font, sharp focus, and high contrast against dark wood", "target_evidence": "text region with 'ROOM 18'", "expected_visual_elements": ["menu board", "white font", "dark wood"], "negative_prompt": "faded text, low resolution, watermarks, logos, real brands", "diversity_tags": ["menu board", "white font", "dark wood"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ed267d7e662f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic menu board with", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.175, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_ed267d7e662f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_ed267d7e662f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22982116, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_ed267d7e662f/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_ed267d7e662f/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_ed267d7e662f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1518663, "qwen_elapsed_seconds": 5.446, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_ed267d7e662f/record.json"} +{"prompt_id": "ocr_heavy_d3_ff963e372bc5", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 3, "image_prompt": "Synthetic photorealistic poster with 'DELTA' in large yellow letters on a dark blue background, sharp focus, clear foreground separation.", "target_evidence": "DELTA", "expected_visual_elements": ["poster", "yellow text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["poster", "yellow text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ff963e372bc5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic poster with delta", "max_similarity_to_other_prompt": 0.6, "diversity_score": 0.2, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d3_ff963e372bc5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d3_ff963e372bc5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29436926, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d3_ff963e372bc5/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_ff963e372bc5/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d3_ff963e372bc5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8044694, "qwen_elapsed_seconds": 5.69, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d3_ff963e372bc5/record.json"} +{"prompt_id": "ocr_heavy_d4_03dc84353225", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic package with 'ORBIT' on a blue sticker, clear text boundaries, and high contrast against a white surface", "target_evidence": "ORBIT", "expected_visual_elements": ["package", "blue sticker", "high contrast"], "negative_prompt": "real brands, watermarks, famous logos, blurred text, low contrast", "diversity_tags": ["package", "blue sticker", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "03dc84353225", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic package with orbit", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_03dc84353225", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_03dc84353225/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29353057, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_03dc84353225/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_03dc84353225/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_03dc84353225/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1424124, "qwen_elapsed_seconds": 5.503, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_03dc84353225/record.json"} +{"prompt_id": "ocr_heavy_d4_1020e6f64244", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic poster with 'TAXI 5' in bold yellow text on a dark background, clear foreground separation, and high contrast", "target_evidence": "TAXI 5", "expected_visual_elements": ["poster", "bold yellow text", "dark background"], "negative_prompt": "real brands, watermarks, famous logos, blurred text, low contrast", "diversity_tags": ["poster", "bold yellow", "dark background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1020e6f64244", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic poster with taxi", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_1020e6f64244", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_1020e6f64244/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29304196, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_1020e6f64244/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_1020e6f64244/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_1020e6f64244/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3266486, "qwen_elapsed_seconds": 5.73, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_1020e6f64244/record.json"} +{"prompt_id": "ocr_heavy_d4_13202b52ae27", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic screen displaying 'DELTA' with high contrast text, clean background, and clear visual separation from surrounding elements", "target_evidence": "DELTA", "expected_visual_elements": ["screen", "high contrast text", "clean background"], "negative_prompt": "real brands, watermarks, famous logos, blurred text, low contrast", "diversity_tags": ["screen", "high contrast", "clean background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "13202b52ae27", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying delta", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_13202b52ae27", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_13202b52ae27/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29710134, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_13202b52ae27/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_13202b52ae27/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_13202b52ae27/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9197868, "qwen_elapsed_seconds": 5.796, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_13202b52ae27/record.json"} +{"prompt_id": "ocr_heavy_d4_22c579c51775", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic sign with yellow text 'EXIT B' on a dark green board, clear boundaries and high contrast", "target_evidence": "EXIT B", "expected_visual_elements": ["sign", "yellow text", "dark green board"], "negative_prompt": "low contrast, blurry text, real logos, extra details", "diversity_tags": ["sign", "yellow text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "22c579c51775", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic sign with yellow", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_22c579c51775", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_22c579c51775/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29276689, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_22c579c51775/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_22c579c51775/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_22c579c51775/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2684643, "qwen_elapsed_seconds": 5.96, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_22c579c51775/record.json"} +{"prompt_id": "ocr_heavy_d4_282d36c34e17", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic jersey with 'KILO' printed in black on a white sleeve, clean boundaries, and clear visual separation from the rest of the design", "target_evidence": "KILO", "expected_visual_elements": ["jersey", "white sleeve", "clean boundaries"], "negative_prompt": "real brands, watermarks, famous logos, blurred text, low contrast", "diversity_tags": ["jersey", "white sleeve", "clean boundaries"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "282d36c34e17", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic jersey with kilo", "max_similarity_to_other_prompt": 0.4333, "diversity_score": 0.5667, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d4_282d36c34e17", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_282d36c34e17/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29343787, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_282d36c34e17/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_282d36c34e17/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_282d36c34e17/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 629603, "qwen_elapsed_seconds": 5.65, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_282d36c34e17/record.json"} +{"prompt_id": "ocr_heavy_d4_30ac0a8f6dcd", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic jersey with 'LIMA' in green on a black background, clear text separation and sharp focus on the front panel", "target_evidence": "text 'LIMA' on jersey", "expected_visual_elements": ["jersey", "black background", "green text"], "negative_prompt": "low resolution, watermarks, real brands, famous logos", "diversity_tags": ["jersey", "green text", "black background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "30ac0a8f6dcd", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic jersey with lima", "max_similarity_to_other_prompt": 0.5833, "diversity_score": 0.4167, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d4_30ac0a8f6dcd", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_30ac0a8f6dcd/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22646035, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_30ac0a8f6dcd/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_30ac0a8f6dcd/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_30ac0a8f6dcd/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4284627, "qwen_elapsed_seconds": 5.692, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_30ac0a8f6dcd/record.json"} +{"prompt_id": "ocr_heavy_d4_38ad4072f241", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic jersey with 'ALTO' in white on a dark blue background, clear text separation and sharp focus", "target_evidence": "text 'ALTO' on jersey", "expected_visual_elements": ["jersey", "dark blue background", "white text"], "negative_prompt": "low resolution, watermarks, real brands, famous logos", "diversity_tags": ["jersey", "white text", "dark background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "38ad4072f241", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic jersey with alto", "max_similarity_to_other_prompt": 0.6957, "diversity_score": 0.1043, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_38ad4072f241", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_38ad4072f241/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26869250, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_38ad4072f241/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_38ad4072f241/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_38ad4072f241/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1451702, "qwen_elapsed_seconds": 5.67, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_38ad4072f241/record.json"} +{"prompt_id": "ocr_heavy_d4_403c826e138e", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic crate with 'MESA' label on a red tag, clear text boundaries, and sharp focus against a neutral background", "target_evidence": "MESA", "expected_visual_elements": ["crate", "red tag", "sharp focus"], "negative_prompt": "real brands, watermarks, famous logos, blurred text, low contrast", "diversity_tags": ["crate", "red tag", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "403c826e138e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with mesa", "max_similarity_to_other_prompt": 0.6087, "diversity_score": 0.1913, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_403c826e138e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_403c826e138e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28869650, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_403c826e138e/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_403c826e138e/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_403c826e138e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2697644, "qwen_elapsed_seconds": 5.63, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_403c826e138e/record.json"} +{"prompt_id": "ocr_heavy_d4_49b89a1a0230", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic notebook page with clean text 'UNIT 12' in bold black font on a wooden desk under bright lighting, sharp focus on the text area", "target_evidence": "text 'UNIT 12' on notebook page", "expected_visual_elements": ["wooden desk", "notebook page", "bold black font"], "negative_prompt": "blurry text, low contrast, watermarks, real brands, famous logos", "diversity_tags": ["notebook", "bold font", "wooden desk"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "49b89a1a0230", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.6429, "diversity_score": 0.1571, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_49b89a1a0230", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_49b89a1a0230/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26905525, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_49b89a1a0230/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_49b89a1a0230/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_49b89a1a0230/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1718131, "qwen_elapsed_seconds": 5.709, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_49b89a1a0230/record.json"} +{"prompt_id": "ocr_heavy_d4_548673e556be", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic tag on a crate with 'ZEN' in red on a yellow background, sharp focus and clean text boundaries", "target_evidence": "text 'ZEN' on tag", "expected_visual_elements": ["crate", "yellow background", "red text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands", "diversity_tags": ["tag", "red text", "yellow background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "548673e556be", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tag on a", "max_similarity_to_other_prompt": 0.6087, "diversity_score": 0.1913, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_548673e556be", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_548673e556be/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27273786, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_548673e556be/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_548673e556be/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_548673e556be/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5893944, "qwen_elapsed_seconds": 5.759, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_548673e556be/record.json"} +{"prompt_id": "ocr_heavy_d4_55dfb40f7dab", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic menu board with light blue text 'DELTA' on a wooden panel, sharp focus and clear foreground separation", "target_evidence": "DELTA", "expected_visual_elements": ["menu board", "light blue text", "wooden panel"], "negative_prompt": "faded text, poor lighting, real logos, extra elements", "diversity_tags": ["menu board", "light blue text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "55dfb40f7dab", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic menu board with", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_55dfb40f7dab", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_55dfb40f7dab/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27072687, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_55dfb40f7dab/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_55dfb40f7dab/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_55dfb40f7dab/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1162349, "qwen_elapsed_seconds": 5.636, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_55dfb40f7dab/record.json"} +{"prompt_id": "ocr_heavy_d4_5a9d9859408f", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic menu board with 'ZEN' on a green sign, clean text region, and sharp focus against a neutral background", "target_evidence": "ZEN", "expected_visual_elements": ["menu board", "green sign", "sharp focus"], "negative_prompt": "real brands, watermarks, famous logos, blurred text, low contrast", "diversity_tags": ["menu board", "green sign", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5a9d9859408f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic menu board with", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_5a9d9859408f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_5a9d9859408f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21937122, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_5a9d9859408f/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_5a9d9859408f/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_5a9d9859408f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8043279, "qwen_elapsed_seconds": 5.568, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_5a9d9859408f/record.json"} +{"prompt_id": "ocr_heavy_d4_6d4dcb3c9538", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic screen displaying 'DELTA' with high contrast white text on dark blue background, clear foreground separation and sharp edges", "target_evidence": "text 'DELTA' on screen", "expected_visual_elements": ["screen", "dark blue background", "white text"], "negative_prompt": "low resolution, watermarks, real brands, famous logos", "diversity_tags": ["screen", "high contrast", "dark background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6d4dcb3c9538", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying delta", "max_similarity_to_other_prompt": 0.56, "diversity_score": 0.24, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_6d4dcb3c9538", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_6d4dcb3c9538/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26668339, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_6d4dcb3c9538/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_6d4dcb3c9538/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_6d4dcb3c9538/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9492746, "qwen_elapsed_seconds": 5.903, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_6d4dcb3c9538/record.json"} +{"prompt_id": "ocr_heavy_d4_78bd4cbf4e31", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic crate with bright red label 'TAXI 5' on a white background, high contrast and clean boundaries", "target_evidence": "TAXI 5", "expected_visual_elements": ["crate", "red label", "white background"], "negative_prompt": "faded text, overlapping elements, real logos, low resolution", "diversity_tags": ["crate", "red label"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "78bd4cbf4e31", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with bright", "max_similarity_to_other_prompt": 0.7143, "diversity_score": 0.0857, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_78bd4cbf4e31", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_78bd4cbf4e31/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24421677, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_78bd4cbf4e31/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_78bd4cbf4e31/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_78bd4cbf4e31/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3949601, "qwen_elapsed_seconds": 5.641, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_78bd4cbf4e31/record.json"} +{"prompt_id": "ocr_heavy_d4_79e5fc8edf08", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic notebook page with clean white text 'UNIT 12' centered on a dark blue cover, sharp focus and clear foreground separation", "target_evidence": "UNIT 12", "expected_visual_elements": ["notebook", "white text", "dark blue cover"], "negative_prompt": "blurry text, low contrast, real brand logos, watermarks, extra elements", "diversity_tags": ["notebook", "centered text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "79e5fc8edf08", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.68, "diversity_score": 0.12, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_79e5fc8edf08", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_79e5fc8edf08/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21110403, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_79e5fc8edf08/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_79e5fc8edf08/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_79e5fc8edf08/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1407896, "qwen_elapsed_seconds": 5.528, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_79e5fc8edf08/record.json"} +{"prompt_id": "ocr_heavy_d4_83f557e64352", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic tag on a black crate with red text 'GATE 42', sharp focus and clear foreground separation", "target_evidence": "GATE 42", "expected_visual_elements": ["tag", "red text", "black crate"], "negative_prompt": "faded text, poor lighting, real brands, extra elements", "diversity_tags": ["tag", "red text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "83f557e64352", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tag on a", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_83f557e64352", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_83f557e64352/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27187391, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_83f557e64352/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_83f557e64352/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_83f557e64352/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1585625, "qwen_elapsed_seconds": 5.599, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_83f557e64352/record.json"} +{"prompt_id": "ocr_heavy_d4_966ad181518c", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic menu board with 'RIVER 9' in blue on a light gray background, sharp focus and clean text boundaries", "target_evidence": "text 'RIVER 9' on menu board", "expected_visual_elements": ["menu board", "light gray background", "blue text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands", "diversity_tags": ["menu board", "blue text", "light background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "966ad181518c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic menu board with", "max_similarity_to_other_prompt": 0.5556, "diversity_score": 0.2444, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_966ad181518c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_966ad181518c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25348331, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_966ad181518c/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_966ad181518c/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_966ad181518c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2228028, "qwen_elapsed_seconds": 5.464, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_966ad181518c/record.json"} +{"prompt_id": "ocr_heavy_d4_ba56b0c836bc", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic crate with 'MESA' printed in red on a white label, clean boundaries and sharp focus on the label area", "target_evidence": "text 'MESA' on crate label", "expected_visual_elements": ["crate", "white label", "red text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands", "diversity_tags": ["crate", "white label", "red text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ba56b0c836bc", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with mesa", "max_similarity_to_other_prompt": 0.6071, "diversity_score": 0.1929, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_ba56b0c836bc", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_ba56b0c836bc/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20427251, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_ba56b0c836bc/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_ba56b0c836bc/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_ba56b0c836bc/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8901079, "qwen_elapsed_seconds": 5.617, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_ba56b0c836bc/record.json"} +{"prompt_id": "ocr_heavy_d4_c5032f8dd4cb", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic notebook page with clean text region 'UNIT 12' on a white border, sharp focus, clear foreground separation, and no background clutter", "target_evidence": "UNIT 12", "expected_visual_elements": ["notebook page", "white border", "sharp focus"], "negative_prompt": "real brands, watermarks, famous logos, blurred text, low contrast", "diversity_tags": ["notebook", "white border", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c5032f8dd4cb", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.6923, "diversity_score": 0.1077, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_c5032f8dd4cb", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_c5032f8dd4cb/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21646678, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_c5032f8dd4cb/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_c5032f8dd4cb/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_c5032f8dd4cb/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8510605, "qwen_elapsed_seconds": 5.722, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_c5032f8dd4cb/record.json"} +{"prompt_id": "ocr_heavy_d4_d1449a26e5b0", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 4, "image_prompt": "synthetic photorealistic poster with 'TAXI 5' in yellow on a dark green background, sharp focus on the text area and clean boundaries", "target_evidence": "text 'TAXI 5' on poster", "expected_visual_elements": ["poster", "dark green background", "yellow text"], "negative_prompt": "blurry text, low contrast, watermarks, real brands", "diversity_tags": ["poster", "yellow text", "dark background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d1449a26e5b0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic poster with taxi", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.175, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d4_d1449a26e5b0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d4_d1449a26e5b0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24144531, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d4_d1449a26e5b0/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_d1449a26e5b0/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d4_d1449a26e5b0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1207071, "qwen_elapsed_seconds": 5.773, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d4_d1449a26e5b0/record.json"} +{"prompt_id": "ocr_heavy_d5_077cf55a1dd7", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic crate with sharp focus and clear foreground, text 'TAXI 5' in white on a dark blue label, stacked on the ground", "target_evidence": "text 'TAXI 5' on crate label", "expected_visual_elements": ["crate", "label"], "negative_prompt": "faded text, low contrast, real brand logos, watermarks", "diversity_tags": ["crate", "label"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "077cf55a1dd7", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with sharp", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.4615, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d5_077cf55a1dd7", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_077cf55a1dd7/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27253524, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_077cf55a1dd7/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_077cf55a1dd7/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_077cf55a1dd7/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7626443, "qwen_elapsed_seconds": 5.944, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_077cf55a1dd7/record.json"} +{"prompt_id": "ocr_heavy_d5_15a06334fbe4", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic package with high contrast and clean boundaries, text 'GATE 42' in black on a white label, stacked on a shelf", "target_evidence": "text 'GATE 42' on package", "expected_visual_elements": ["package", "label"], "negative_prompt": "blurry text, low contrast, real brand logos, watermarks", "diversity_tags": ["package", "shelf"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "15a06334fbe4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic package with high", "max_similarity_to_other_prompt": 0.6957, "diversity_score": 0.3043, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d5_15a06334fbe4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_15a06334fbe4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21014559, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_15a06334fbe4/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_15a06334fbe4/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_15a06334fbe4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8555870, "qwen_elapsed_seconds": 5.536, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_15a06334fbe4/record.json"} +{"prompt_id": "ocr_heavy_d5_1a5b4fac2177", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic poster with 'ZEN' in bold, high contrast letters on a dark background, clean and sharp design", "target_evidence": "text 'ZEN' on poster", "expected_visual_elements": ["poster", "dark background", "bold text"], "negative_prompt": "faded text, cluttered background, real logos, distorted text", "diversity_tags": ["poster", "dark background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1a5b4fac2177", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic poster with zen", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.4348, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d5_1a5b4fac2177", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_1a5b4fac2177/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22131975, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_1a5b4fac2177/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_1a5b4fac2177/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_1a5b4fac2177/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3372794, "qwen_elapsed_seconds": 5.616, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_1a5b4fac2177/record.json"} +{"prompt_id": "ocr_heavy_d5_37c514cdd8e2", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic sign with 'GATE 42' in a clean, high-contrast font on a white board, clear foreground separation", "target_evidence": "text 'GATE 42' on sign", "expected_visual_elements": ["sign", "white background", "clean font"], "negative_prompt": "faded text, cluttered background, real logos, low contrast", "diversity_tags": ["sign", "white background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "37c514cdd8e2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic sign with gate", "max_similarity_to_other_prompt": 0.5217, "diversity_score": 0.4783, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d5_37c514cdd8e2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_37c514cdd8e2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27344877, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_37c514cdd8e2/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_37c514cdd8e2/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_37c514cdd8e2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8715621, "qwen_elapsed_seconds": 6.17, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_37c514cdd8e2/record.json"} +{"prompt_id": "ocr_heavy_d5_454a6be8e58d", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic notebook page with clear foreground separation, containing the text 'UNIT 12' in bold black font on a white background, surrounded by clean lines and sharp focus", "target_evidence": "text region with 'UNIT 12'", "expected_visual_elements": ["notebook page", "bold black text", "white background"], "negative_prompt": "blurry text, low contrast, real brands, watermarks, celebrities, famous logos", "diversity_tags": ["notebook", "bold font", "white background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "454a6be8e58d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.75, "diversity_score": 0.05, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d5_454a6be8e58d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_454a6be8e58d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22820619, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_454a6be8e58d/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_454a6be8e58d/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_454a6be8e58d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1015754, "qwen_elapsed_seconds": 5.551, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_454a6be8e58d/record.json"} +{"prompt_id": "ocr_heavy_d5_5278b7ef8c6a", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic screen displaying sharp focus and clear foreground, text 'ORBIT' in white on a dark background, part of a dashboard", "target_evidence": "text 'ORBIT' on screen", "expected_visual_elements": ["screen", "dashboard"], "negative_prompt": "faded text, low contrast, real brand logos, watermarks", "diversity_tags": ["screen", "dashboard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5278b7ef8c6a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying sharp", "max_similarity_to_other_prompt": 0.6, "diversity_score": 0.2, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d5_5278b7ef8c6a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_5278b7ef8c6a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22310793, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_5278b7ef8c6a/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_5278b7ef8c6a/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_5278b7ef8c6a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 245701, "qwen_elapsed_seconds": 5.971, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_5278b7ef8c6a/record.json"} +{"prompt_id": "ocr_heavy_d5_549f7f9da427", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic notebook page with 'BLUE-7' written in pencil, visible on a clean white sheet, sharp focus", "target_evidence": "BLUE-7", "expected_visual_elements": ["notebook page", "pencil writing", "clean white sheet"], "negative_prompt": "smudged text, real brands, watermarks, blurred edges", "diversity_tags": ["notebook page", "pencil writing", "clean white sheet"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "549f7f9da427", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d5_549f7f9da427", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_549f7f9da427/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20324232, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_549f7f9da427/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_549f7f9da427/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_549f7f9da427/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7959168, "qwen_elapsed_seconds": 5.874, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_549f7f9da427/record.json"} +{"prompt_id": "ocr_heavy_d5_68980ef803ab", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic crate with high contrast lighting, featuring the text 'TAXI 5' on a sticker with clean boundaries and sharp focus, placed near a metal tool", "target_evidence": "text region with 'TAXI 5'", "expected_visual_elements": ["crate", "sticker", "high contrast lighting", "metal tool"], "negative_prompt": "blurry text, low contrast, real brands, watermarks, celebrities, famous logos", "diversity_tags": ["crate", "sticker", "metal tool"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "68980ef803ab", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with high", "max_similarity_to_other_prompt": 0.6154, "diversity_score": 0.3846, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d5_68980ef803ab", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_68980ef803ab/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21726815, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_68980ef803ab/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_68980ef803ab/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_68980ef803ab/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7285712, "qwen_elapsed_seconds": 5.892, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_68980ef803ab/record.json"} +{"prompt_id": "ocr_heavy_d5_74fc1c35c473", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic crate with 'RIVER 9' printed on a label, clear boundaries, synthetic lighting, visible from the side", "target_evidence": "RIVER 9", "expected_visual_elements": ["crate", "label", "synthetic lighting", "side view"], "negative_prompt": "faded text, real brands, watermarks, low resolution", "diversity_tags": ["crate", "label", "side view"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "74fc1c35c473", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic crate with river", "max_similarity_to_other_prompt": 0.4, "diversity_score": 0.4, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d5_74fc1c35c473", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_74fc1c35c473/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29727816, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_74fc1c35c473/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_74fc1c35c473/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_74fc1c35c473/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2799872, "qwen_elapsed_seconds": 5.481, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_74fc1c35c473/record.json"} +{"prompt_id": "ocr_heavy_d5_9c3ac4f01e0a", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic screen displaying 'UNIT 12' with high contrast text on a dark background, clean boundaries, sharp focus", "target_evidence": "UNIT 12", "expected_visual_elements": ["screen", "dark background", "high contrast text"], "negative_prompt": "blurry text, low contrast, real brands, watermarks, celebrities", "diversity_tags": ["screen", "dark background", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "9c3ac4f01e0a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying unit", "max_similarity_to_other_prompt": 1.0, "diversity_score": 0.0, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d5_9c3ac4f01e0a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_9c3ac4f01e0a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21934490, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_9c3ac4f01e0a/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_9c3ac4f01e0a/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_9c3ac4f01e0a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 73130, "qwen_elapsed_seconds": 5.911, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_9c3ac4f01e0a/record.json"} +{"prompt_id": "ocr_heavy_d5_a1d7ea50f51f", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic menu board with clean boundaries and high contrast, text 'MESA' in yellow on a black background, displayed in a restaurant", "target_evidence": "text 'MESA' on menu board", "expected_visual_elements": ["menu board", "yellow text"], "negative_prompt": "blurry text, low contrast, real brand logos, watermarks", "diversity_tags": ["menu board", "restaurant"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a1d7ea50f51f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic menu board with", "max_similarity_to_other_prompt": 0.5185, "diversity_score": 0.2815, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d5_a1d7ea50f51f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_a1d7ea50f51f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27988147, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_a1d7ea50f51f/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_a1d7ea50f51f/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_a1d7ea50f51f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 668245, "qwen_elapsed_seconds": 5.583, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_a1d7ea50f51f/record.json"} +{"prompt_id": "ocr_heavy_d5_a9deb5ced6e0", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic screen displaying sharp focus and clear foreground, text 'LIMA' in green on a dark background, embedded in a control panel", "target_evidence": "text 'LIMA' on screen", "expected_visual_elements": ["screen", "control panel"], "negative_prompt": "faded text, low contrast, real brand logos, watermarks", "diversity_tags": ["screen", "control panel"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a9deb5ced6e0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic screen displaying sharp", "max_similarity_to_other_prompt": 0.6, "diversity_score": 0.2, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d5_a9deb5ced6e0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_a9deb5ced6e0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23580959, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_a9deb5ced6e0/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_a9deb5ced6e0/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_a9deb5ced6e0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4249353, "qwen_elapsed_seconds": 5.561, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_a9deb5ced6e0/record.json"} +{"prompt_id": "ocr_heavy_d5_aa56e3c6a137", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic sign with sharp focus and clear foreground separation, text 'DELTA' in red on a dark background, mounted on a wall", "target_evidence": "text 'DELTA' on sign", "expected_visual_elements": ["sign", "red text"], "negative_prompt": "faded text, real brand logos, watermarks, cluttered background", "diversity_tags": ["sign", "wall-mounted"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "aa56e3c6a137", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic sign with sharp", "max_similarity_to_other_prompt": 0.64, "diversity_score": 0.36, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d5_aa56e3c6a137", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_aa56e3c6a137/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27750557, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_aa56e3c6a137/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_aa56e3c6a137/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_aa56e3c6a137/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 950346, "qwen_elapsed_seconds": 5.762, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_aa56e3c6a137/record.json"} +{"prompt_id": "ocr_heavy_d5_b6b750f3837b", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic jersey with a label 'TAXI 5' in a contrasting color, clean stitching, and a sports fabric texture", "target_evidence": "label 'TAXI 5' on jersey", "expected_visual_elements": ["jersey", "label", "sports fabric"], "negative_prompt": "low contrast text, real logos, watermarks, distorted stitching", "diversity_tags": ["jersey", "label"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b6b750f3837b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic jersey with a", "max_similarity_to_other_prompt": 0.3571, "diversity_score": 0.6429, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d5_b6b750f3837b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_b6b750f3837b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26958848, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_b6b750f3837b/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_b6b750f3837b/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_b6b750f3837b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1676248, "qwen_elapsed_seconds": 5.7, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_b6b750f3837b/record.json"} +{"prompt_id": "ocr_heavy_d5_be8ea0c7f8f5", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic notebook page with clean boundaries and high contrast, text 'UNIT 12' centered at the top, white background, sharp focus", "target_evidence": "text 'UNIT 12' on notebook page", "expected_visual_elements": ["notebook page", "centered text"], "negative_prompt": "blurry text, low contrast, real brand logos, watermarks", "diversity_tags": ["notebook", "centered text"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "be8ea0c7f8f5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.7083, "diversity_score": 0.0917, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d5_be8ea0c7f8f5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_be8ea0c7f8f5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27895355, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_be8ea0c7f8f5/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_be8ea0c7f8f5/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_be8ea0c7f8f5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6620629, "qwen_elapsed_seconds": 5.58, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_be8ea0c7f8f5/record.json"} +{"prompt_id": "ocr_heavy_d5_c6a2d9fda628", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic notebook page with clear boundaries, a small text region 'ROOM 18' on a clean white background, sharp focus on the text", "target_evidence": "text region with 'ROOM 18'", "expected_visual_elements": ["notebook page", "white background", "sharp text focus"], "negative_prompt": "blurry text, real brand logos, watermarks, complex background", "diversity_tags": ["notebook", "white background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c6a2d9fda628", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notebook page with", "max_similarity_to_other_prompt": 0.6, "diversity_score": 0.2, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d5_c6a2d9fda628", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_c6a2d9fda628/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23749804, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_c6a2d9fda628/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_c6a2d9fda628/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_c6a2d9fda628/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5328250, "qwen_elapsed_seconds": 5.627, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_c6a2d9fda628/record.json"} +{"prompt_id": "ocr_heavy_d5_d35a0d0f21f2", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic menu board with 'ALTO' in a clean, high-contrast font on a white board, sharp focus on text", "target_evidence": "text 'ALTO' on menu board", "expected_visual_elements": ["menu board", "white background", "sharp text"], "negative_prompt": "faded text, cluttered background, real logos, low contrast", "diversity_tags": ["menu board", "white background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d35a0d0f21f2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic menu board with", "max_similarity_to_other_prompt": 0.6087, "diversity_score": 0.1913, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d5_d35a0d0f21f2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_d35a0d0f21f2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29194225, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_d35a0d0f21f2/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_d35a0d0f21f2/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_d35a0d0f21f2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5024551, "qwen_elapsed_seconds": 5.53, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_d35a0d0f21f2/record.json"} +{"prompt_id": "ocr_heavy_d5_db75b0748b7e", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic jersey with high contrast and clean boundaries, text 'ALTO' in black on a light blue background, worn fabric texture", "target_evidence": "text 'ALTO' on jersey", "expected_visual_elements": ["jersey", "text on fabric"], "negative_prompt": "low contrast, blurry text, real logos, watermarks", "diversity_tags": ["jersey", "fabric texture"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "db75b0748b7e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic jersey with high", "max_similarity_to_other_prompt": 0.5185, "diversity_score": 0.4815, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d5_db75b0748b7e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_db75b0748b7e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26656383, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_db75b0748b7e/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_db75b0748b7e/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_db75b0748b7e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3658150, "qwen_elapsed_seconds": 6.182, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_db75b0748b7e/record.json"} +{"prompt_id": "ocr_heavy_d5_f5a708c509ca", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic poster with sharp focus and clear foreground separation, text 'ZEN' in white on a dark blue background, hanging on a wall", "target_evidence": "text 'ZEN' on poster", "expected_visual_elements": ["poster", "wall-mounted"], "negative_prompt": "faded text, low contrast, real brand logos, watermarks", "diversity_tags": ["poster", "wall-mounted"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f5a708c509ca", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic poster with sharp", "max_similarity_to_other_prompt": 0.6957, "diversity_score": 0.3043, "repeated_opening": false, "review": {"prompt_id": "ocr_heavy_d5_f5a708c509ca", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_f5a708c509ca/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29901775, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_f5a708c509ca/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_f5a708c509ca/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_f5a708c509ca/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9455644, "qwen_elapsed_seconds": 5.613, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_f5a708c509ca/record.json"} +{"prompt_id": "ocr_heavy_d5_f75d44bf94ac", "run_name": "flux2_qil_decomposition_pilot", "category": "ocr_heavy", "difficulty": 5, "image_prompt": "synthetic photorealistic tag on a clean background with high contrast, text 'RIVER 9' in black on a light yellow label, attached to a box", "target_evidence": "text 'RIVER 9' on tag", "expected_visual_elements": ["tag", "label"], "negative_prompt": "blurry text, low contrast, real brand logos, watermarks", "diversity_tags": ["tag", "box"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f75d44bf94ac", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic tag on a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "ocr_heavy_d5_f75d44bf94ac", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/ocr_heavy/ocr_heavy_d5_f75d44bf94ac/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27626645, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/ocr_heavy/ocr_heavy_d5_f75d44bf94ac/layers/layer_00.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_f75d44bf94ac/layers/layer_01.png", "layers-x/k3/ocr_heavy/ocr_heavy_d5_f75d44bf94ac/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4562191, "qwen_elapsed_seconds": 5.588, "qwen_record_path": "layers-x/k3/ocr_heavy/ocr_heavy_d5_f75d44bf94ac/record.json"} +{"prompt_id": "single_evidence_control_d1_10b901f87f97", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "synthetic photorealistic low angle shot of a single black leather wallet on a wooden desk, sharp focus and clear foreground separation", "target_evidence": "black leather wallet", "expected_visual_elements": ["leather wallet", "black color", "wooden desk"], "negative_prompt": "distorted shapes, reflections, shadows, multiple objects", "diversity_tags": ["color", "material", "background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "10b901f87f97", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic low angle shot", "max_similarity_to_other_prompt": 0.5185, "diversity_score": 0.2815, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_10b901f87f97", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_10b901f87f97/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28122394, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_10b901f87f97/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_10b901f87f97/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_10b901f87f97/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5770233, "qwen_elapsed_seconds": 6.21, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_10b901f87f97/record.json"} +{"prompt_id": "single_evidence_control_d1_1cf5d97c3735", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "Synthetic photorealistic macro view of a textured brown wooden block with sharp focus and clean boundaries, displaying the block's natural grain and high contrast against the white background", "target_evidence": "one obvious attribute such as color, material, shape, or pattern", "expected_visual_elements": ["textured brown wooden block", "white background"], "negative_prompt": "distractors, soft focus, multiple objects, complex lighting", "diversity_tags": ["wood", "textured", "macro"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1cf5d97c3735", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic macro view of", "max_similarity_to_other_prompt": 0.5152, "diversity_score": 0.2848, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_1cf5d97c3735", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_1cf5d97c3735/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27946754, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_1cf5d97c3735/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_1cf5d97c3735/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_1cf5d97c3735/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5767631, "qwen_elapsed_seconds": 5.71, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_1cf5d97c3735/record.json"} +{"prompt_id": "single_evidence_control_d1_422179b0a73b", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "Synthetic photorealistic side view of a smooth black rubber ball on a light gray surface with clean boundaries and high contrast, emphasizing the ball's shape and material", "target_evidence": "one obvious attribute such as color, material, shape, or pattern", "expected_visual_elements": ["smooth black rubber ball", "light gray surface"], "negative_prompt": "distractors, soft focus, multiple objects, complex lighting", "diversity_tags": ["ball", "rubber", "side view"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "422179b0a73b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic side view of", "max_similarity_to_other_prompt": 0.4516, "diversity_score": 0.3484, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_422179b0a73b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_422179b0a73b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22175260, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_422179b0a73b/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_422179b0a73b/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_422179b0a73b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 215032, "qwen_elapsed_seconds": 5.948, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_422179b0a73b/record.json"} +{"prompt_id": "single_evidence_control_d1_43156b103f60", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "synthetic photorealistic close-up of a single green rubber band on a light gray surface, high contrast and clear foreground separation", "target_evidence": "green rubber band", "expected_visual_elements": ["rubber band", "green color", "light gray surface"], "negative_prompt": "distorted shapes, shadows, reflections, complex textures", "diversity_tags": ["color", "material", "surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "43156b103f60", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_43156b103f60", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_43156b103f60/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21357990, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_43156b103f60/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_43156b103f60/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_43156b103f60/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1761108, "qwen_elapsed_seconds": 5.749, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_43156b103f60/record.json"} +{"prompt_id": "single_evidence_control_d1_781421c6b7cf", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "Synthetic photorealistic overhead shot of a bright orange plastic bowl on a black tile floor with clear separation and high contrast, emphasizing the bowl's color and shape", "target_evidence": "one central target object", "expected_visual_elements": ["bright orange plastic bowl", "black tile floor"], "negative_prompt": "low contrast, complex background, multiple objects, blurred edges", "diversity_tags": ["bowl", "plastic", "overhead"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "781421c6b7cf", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic overhead shot of", "max_similarity_to_other_prompt": 0.5455, "diversity_score": 0.2545, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_781421c6b7cf", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_781421c6b7cf/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23125283, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_781421c6b7cf/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_781421c6b7cf/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_781421c6b7cf/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1437115, "qwen_elapsed_seconds": 5.771, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_781421c6b7cf/record.json"} +{"prompt_id": "single_evidence_control_d1_814450e25557", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "synthetic photorealistic side view of a single white plastic cup with a clear lid on a blue fabric surface, high contrast and clean boundaries", "target_evidence": "white plastic cup", "expected_visual_elements": ["plastic cup", "white color", "blue fabric surface"], "negative_prompt": "blurry, reflections, shadows, complex patterns", "diversity_tags": ["color", "material", "surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "814450e25557", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic side view of", "max_similarity_to_other_prompt": 0.5172, "diversity_score": 0.2828, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_814450e25557", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_814450e25557/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23975564, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_814450e25557/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_814450e25557/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_814450e25557/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1309212, "qwen_elapsed_seconds": 6.41, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_814450e25557/record.json"} +{"prompt_id": "single_evidence_control_d1_9278a977aeb8", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "synthetic photorealistic close-up of a smooth ceramic bowl with glossy surface on a white matte background, high contrast and clean boundaries", "target_evidence": "smooth ceramic bowl", "expected_visual_elements": ["ceramic bowl", "glossy surface", "white matte background"], "negative_prompt": "blurry, low contrast, complex patterns, multiple objects, shadows", "diversity_tags": ["ceramic", "surface texture", "background material"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "9278a977aeb8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.7083, "diversity_score": 0.0917, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_9278a977aeb8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_9278a977aeb8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20591732, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_9278a977aeb8/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_9278a977aeb8/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_9278a977aeb8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8874389, "qwen_elapsed_seconds": 5.599, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_9278a977aeb8/record.json"} +{"prompt_id": "single_evidence_control_d1_96b13188c2b7", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "synthetic photorealistic wide shot of a wooden cube with a carved pattern on a dark wooden table, high contrast and clear foreground separation", "target_evidence": "carved wooden cube", "expected_visual_elements": ["wooden cube", "carved pattern", "dark wooden table"], "negative_prompt": "low resolution, shadows, reflections, complex backgrounds", "diversity_tags": ["shape", "pattern", "surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "96b13188c2b7", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wide shot of", "max_similarity_to_other_prompt": 0.64, "diversity_score": 0.16, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_96b13188c2b7", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_96b13188c2b7/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27277220, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_96b13188c2b7/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_96b13188c2b7/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_96b13188c2b7/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1801735, "qwen_elapsed_seconds": 5.506, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_96b13188c2b7/record.json"} +{"prompt_id": "single_evidence_control_d1_a92f796e47c1", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "Synthetic photorealistic low-angle shot of a yellow plastic cube on a green felt surface with clear foreground separation and high contrast, emphasizing the cube's color and shape", "target_evidence": "one central target object", "expected_visual_elements": ["yellow plastic cube", "green felt surface"], "negative_prompt": "blurry, low contrast, complex background, multiple objects", "diversity_tags": ["cube", "plastic", "low-angle"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a92f796e47c1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic low angle shot", "max_similarity_to_other_prompt": 0.5455, "diversity_score": 0.2545, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_a92f796e47c1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_a92f796e47c1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24220062, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_a92f796e47c1/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_a92f796e47c1/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_a92f796e47c1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4095605, "qwen_elapsed_seconds": 6.321, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_a92f796e47c1/record.json"} +{"prompt_id": "single_evidence_control_d1_b87076fca0ae", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "Synthetic photorealistic wide shot of a white ceramic mug with sharp focus and clean separation on a dark wooden table, highlighting the mug's shape and high contrast against the background", "target_evidence": "one central target object", "expected_visual_elements": ["white ceramic mug", "dark wooden table"], "negative_prompt": "blurry, low contrast, complex background, multiple objects", "diversity_tags": ["mug", "ceramic", "wide shot"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b87076fca0ae", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wide shot of", "max_similarity_to_other_prompt": 0.5455, "diversity_score": 0.2545, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_b87076fca0ae", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_b87076fca0ae/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25476619, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_b87076fca0ae/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_b87076fca0ae/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_b87076fca0ae/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7111154, "qwen_elapsed_seconds": 5.671, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_b87076fca0ae/record.json"} +{"prompt_id": "single_evidence_control_d1_bd963a816d43", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "Synthetic photorealistic close-up of a patterned red fabric square on a plain white background with clear foreground separation and high contrast, emphasizing the fabric's pattern", "target_evidence": "one obvious attribute such as color, material, shape, or pattern", "expected_visual_elements": ["patterned red fabric square", "plain white background"], "negative_prompt": "distractors, soft focus, multiple objects, complex lighting", "diversity_tags": ["fabric", "patterned", "close-up"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "bd963a816d43", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.4545, "diversity_score": 0.3455, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_bd963a816d43", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_bd963a816d43/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28877733, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_bd963a816d43/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_bd963a816d43/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_bd963a816d43/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7955063, "qwen_elapsed_seconds": 5.651, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_bd963a816d43/record.json"} +{"prompt_id": "single_evidence_control_d1_be571dcc0147", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "Synthetic photorealistic portrait of a smooth white marble statue with sharp focus and clean boundaries, highlighting the statue's smooth surface and high contrast against the dark background", "target_evidence": "one obvious attribute such as color, material, shape, or pattern", "expected_visual_elements": ["smooth white marble statue", "dark background"], "negative_prompt": "distractors, soft focus, multiple objects, complex lighting", "diversity_tags": ["statue", "marble", "smooth"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "be571dcc0147", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic portrait of a", "max_similarity_to_other_prompt": 0.5455, "diversity_score": 0.2545, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_be571dcc0147", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_be571dcc0147/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26052170, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_be571dcc0147/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_be571dcc0147/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_be571dcc0147/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3452943, "qwen_elapsed_seconds": 5.73, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_be571dcc0147/record.json"} +{"prompt_id": "single_evidence_control_d1_c2657d58ae61", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "synthetic photorealistic portrait of a single metallic key with a brushed finish on a light gray surface, clean boundaries and high contrast", "target_evidence": "metallic key", "expected_visual_elements": ["metallic key", "brushed finish", "light gray surface"], "negative_prompt": "low detail, shadows, complex textures, multiple objects", "diversity_tags": ["metallic", "finish", "surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c2657d58ae61", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic portrait of a", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_c2657d58ae61", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_c2657d58ae61/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26431326, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_c2657d58ae61/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_c2657d58ae61/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_c2657d58ae61/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2223218, "qwen_elapsed_seconds": 5.529, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_c2657d58ae61/record.json"} +{"prompt_id": "single_evidence_control_d1_d44193b032da", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "synthetic photorealistic overhead shot of a single blue glass vial with a clear label on a white ceramic surface, sharp focus and clean edges", "target_evidence": "blue glass vial", "expected_visual_elements": ["glass vial", "blue color", "white ceramic surface"], "negative_prompt": "blurry, reflections, shadows, multiple objects", "diversity_tags": ["color", "material", "surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d44193b032da", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic overhead shot of", "max_similarity_to_other_prompt": 0.6154, "diversity_score": 0.1846, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_d44193b032da", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_d44193b032da/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29745547, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_d44193b032da/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_d44193b032da/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_d44193b032da/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2206197, "qwen_elapsed_seconds": 5.613, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_d44193b032da/record.json"} +{"prompt_id": "single_evidence_control_d1_d5a6488be43b", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "synthetic photorealistic overhead shot of a single red rubber ball on a green felt surface with sharp focus and clean separation", "target_evidence": "red rubber ball", "expected_visual_elements": ["rubber ball", "red color", "green felt background"], "negative_prompt": "distorted shapes, uneven lighting, reflections, multiple objects", "diversity_tags": ["color", "material", "background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d5a6488be43b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic overhead shot of", "max_similarity_to_other_prompt": 0.6957, "diversity_score": 0.1043, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_d5a6488be43b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_d5a6488be43b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22930471, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_d5a6488be43b/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_d5a6488be43b/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_d5a6488be43b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3909631, "qwen_elapsed_seconds": 5.859, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_d5a6488be43b/record.json"} +{"prompt_id": "single_evidence_control_d1_db68a0400f05", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "Synthetic photorealistic wide-angle shot of a transparent blue glass cylinder on a wooden table with high contrast and clear foreground separation", "target_evidence": "one central target object", "expected_visual_elements": ["transparent blue glass cylinder", "wooden table"], "negative_prompt": "low contrast, complex background, multiple objects, blurred edges", "diversity_tags": ["glass", "transparent", "wide-angle"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "db68a0400f05", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wide angle shot", "max_similarity_to_other_prompt": 0.64, "diversity_score": 0.16, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_db68a0400f05", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_db68a0400f05/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21224923, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_db68a0400f05/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_db68a0400f05/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_db68a0400f05/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9986093, "qwen_elapsed_seconds": 5.526, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_db68a0400f05/record.json"} +{"prompt_id": "single_evidence_control_d1_e3ea0b79e3b5", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "Synthetic photorealistic close-up of a glossy red apple on a white ceramic plate with clean boundaries and high contrast, sharp focus on the apple's surface texture", "target_evidence": "one central target object", "expected_visual_elements": ["glossy red apple", "white ceramic plate"], "negative_prompt": "blurry, low contrast, complex background, multiple objects", "diversity_tags": ["fruit", "ceramic", "close-up"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e3ea0b79e3b5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.5862, "diversity_score": 0.2138, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_e3ea0b79e3b5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_e3ea0b79e3b5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24240304, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_e3ea0b79e3b5/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_e3ea0b79e3b5/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_e3ea0b79e3b5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 744288, "qwen_elapsed_seconds": 5.674, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_e3ea0b79e3b5/record.json"} +{"prompt_id": "single_evidence_control_d1_e4f77dff0280", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "synthetic photorealistic wide shot of a single silver wireframe cube on a black matte background, clear foreground separation and high contrast", "target_evidence": "silver wireframe cube", "expected_visual_elements": ["wireframe cube", "silver color", "black matte background"], "negative_prompt": "low resolution, shadows, reflections, complex textures", "diversity_tags": ["color", "shape", "background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e4f77dff0280", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wide shot of", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_e4f77dff0280", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_e4f77dff0280/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29360527, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_e4f77dff0280/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_e4f77dff0280/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_e4f77dff0280/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 988994, "qwen_elapsed_seconds": 5.808, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_e4f77dff0280/record.json"} +{"prompt_id": "single_evidence_control_d1_ea16c83ef7f5", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "synthetic photorealistic macro view of a transparent glass sphere with a clear surface on a white textured background, sharp focus and clean edges", "target_evidence": "transparent glass sphere", "expected_visual_elements": ["glass sphere", "transparent material", "white textured background"], "negative_prompt": "blurry, reflections, complex patterns, shadows", "diversity_tags": ["material", "transparency", "background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ea16c83ef7f5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic macro view of", "max_similarity_to_other_prompt": 0.5357, "diversity_score": 0.2643, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_ea16c83ef7f5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_ea16c83ef7f5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23042027, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_ea16c83ef7f5/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_ea16c83ef7f5/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_ea16c83ef7f5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3710030, "qwen_elapsed_seconds": 5.469, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_ea16c83ef7f5/record.json"} +{"prompt_id": "single_evidence_control_d1_f5b3f598913e", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 1, "image_prompt": "Synthetic photorealistic view of a metallic silver sphere resting on a black matte surface with sharp focus and clean separation, emphasizing the sphere's reflective quality", "target_evidence": "one obvious attribute such as color, material, shape, or pattern", "expected_visual_elements": ["metallic silver sphere", "black matte surface"], "negative_prompt": "distractors, soft focus, complex lighting, multiple objects", "diversity_tags": ["sphere", "metallic", "reflective"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f5b3f598913e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4848, "diversity_score": 0.3152, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d1_f5b3f598913e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d1_f5b3f598913e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29217822, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d1_f5b3f598913e/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_f5b3f598913e/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d1_f5b3f598913e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6246503, "qwen_elapsed_seconds": 5.604, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d1_f5b3f598913e/record.json"} +{"prompt_id": "single_evidence_control_d2_1230c87a395b", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "synthetic photorealistic shot of a red ceramic bowl with smooth surface on a white ceramic tray, clean boundaries, high contrast, sharp focus", "target_evidence": "red ceramic bowl", "expected_visual_elements": ["red ceramic bowl", "white ceramic tray", "smooth surface"], "negative_prompt": "distractor objects, blurred edges, low contrast, complex background", "diversity_tags": ["ceramic", "color", "surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1230c87a395b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_1230c87a395b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_1230c87a395b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27034991, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_1230c87a395b/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_1230c87a395b/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_1230c87a395b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3312037, "qwen_elapsed_seconds": 5.635, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_1230c87a395b/record.json"} +{"prompt_id": "single_evidence_control_d2_288f536010cc", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "Synthetic photorealistic scene of a black matte sphere resting on a light gray marble slab with clear foreground separation.", "target_evidence": "black matte sphere", "expected_visual_elements": ["black matte sphere", "light gray marble slab"], "negative_prompt": "low resolution, blurry, distorted, cluttered background, extra objects", "diversity_tags": ["sphere", "matte", "marble"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "288f536010cc", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4333, "diversity_score": 0.3667, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_288f536010cc", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_288f536010cc/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23417738, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_288f536010cc/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_288f536010cc/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_288f536010cc/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9747176, "qwen_elapsed_seconds": 5.827, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_288f536010cc/record.json"} +{"prompt_id": "single_evidence_control_d2_29ebc80ae101", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "synthetic photorealistic shot of a black rubber eraser with a smooth surface on a white plastic background, sharp focus, clean composition", "target_evidence": "black rubber eraser", "expected_visual_elements": ["black rubber eraser", "smooth surface", "white plastic background"], "negative_prompt": "colorful elements, soft shadows, complex background", "diversity_tags": ["material", "color", "surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "29ebc80ae101", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_29ebc80ae101", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_29ebc80ae101/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25618532, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_29ebc80ae101/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_29ebc80ae101/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_29ebc80ae101/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4194784, "qwen_elapsed_seconds": 5.525, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_29ebc80ae101/record.json"} +{"prompt_id": "single_evidence_control_d2_314aa6cf3203", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "synthetic photorealistic close-up of a smooth glass sphere on a dark matte surface, high contrast and sharp focus, clean boundaries, clear foreground separation", "target_evidence": "one central target object", "expected_visual_elements": ["smooth glass sphere", "dark matte surface"], "negative_prompt": "low contrast, blurry edges, multiple objects, complex background", "diversity_tags": ["glass", "matte", "close-up"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "314aa6cf3203", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.7407, "diversity_score": 0.0593, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_314aa6cf3203", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_314aa6cf3203/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23978251, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_314aa6cf3203/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_314aa6cf3203/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_314aa6cf3203/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4593754, "qwen_elapsed_seconds": 5.664, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_314aa6cf3203/record.json"} +{"prompt_id": "single_evidence_control_d2_3f17ccc171fe", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "Synthetic photorealistic shot of a transparent red glass bottle sitting on a dark gray metal tray with sharp focus and clear separation.", "target_evidence": "transparent red glass bottle", "expected_visual_elements": ["transparent red glass bottle", "dark gray metal tray"], "negative_prompt": "low resolution, blurry, distorted, cluttered background, extra objects", "diversity_tags": ["bottle", "glass", "metal"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "3f17ccc171fe", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.4483, "diversity_score": 0.3517, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_3f17ccc171fe", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_3f17ccc171fe/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26227319, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_3f17ccc171fe/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_3f17ccc171fe/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_3f17ccc171fe/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2885881, "qwen_elapsed_seconds": 5.864, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_3f17ccc171fe/record.json"} +{"prompt_id": "single_evidence_control_d2_6417d722a16d", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "synthetic photorealistic image of a green plastic water bottle with a curved shape on a white plastic tray, clear foreground separation", "target_evidence": "green plastic water bottle", "expected_visual_elements": ["green plastic water bottle", "white plastic tray", "curved shape"], "negative_prompt": "colorful elements, soft focus, complex background", "diversity_tags": ["shape", "material", "color"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6417d722a16d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.4138, "diversity_score": 0.3862, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_6417d722a16d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_6417d722a16d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28736595, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_6417d722a16d/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_6417d722a16d/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_6417d722a16d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7437503, "qwen_elapsed_seconds": 5.49, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_6417d722a16d/record.json"} +{"prompt_id": "single_evidence_control_d2_6ab61a3da582", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "synthetic photorealistic shot of a white metal key with a blue handle on a black rubber mat, clean boundaries, clear foreground", "target_evidence": "white metal key", "expected_visual_elements": ["white metal key", "blue handle", "black rubber mat"], "negative_prompt": "colorful elements, soft shadows, complex background", "diversity_tags": ["color", "material", "shape"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6ab61a3da582", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.4074, "diversity_score": 0.3926, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_6ab61a3da582", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_6ab61a3da582/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21738100, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_6ab61a3da582/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_6ab61a3da582/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_6ab61a3da582/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2129877, "qwen_elapsed_seconds": 5.667, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_6ab61a3da582/record.json"} +{"prompt_id": "single_evidence_control_d2_6e293f47d6c3", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "synthetic photorealistic view of a silver metallic spoon on a dark green fabric, high contrast, sharp focus, clean composition", "target_evidence": "silver metallic spoon", "expected_visual_elements": ["silver metallic spoon", "dark green fabric", "high contrast"], "negative_prompt": "distractor objects, blurry edges, complex background", "diversity_tags": ["metallic", "color", "surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6e293f47d6c3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_6e293f47d6c3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_6e293f47d6c3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23548004, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_6e293f47d6c3/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_6e293f47d6c3/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_6e293f47d6c3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5180598, "qwen_elapsed_seconds": 5.477, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_6e293f47d6c3/record.json"} +{"prompt_id": "single_evidence_control_d2_acd46dd75227", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "Synthetic photorealistic close-up of a smooth blue glass sphere on a white matte surface with high contrast and clean boundaries.", "target_evidence": "blue glass sphere", "expected_visual_elements": ["smooth blue glass sphere", "white matte surface"], "negative_prompt": "low resolution, blurry, distorted, cluttered background, extra objects", "diversity_tags": ["glass", "sphere", "close-up"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "acd46dd75227", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_acd46dd75227", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_acd46dd75227/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 30002006, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_acd46dd75227/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_acd46dd75227/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_acd46dd75227/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2688162, "qwen_elapsed_seconds": 5.556, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_acd46dd75227/record.json"} +{"prompt_id": "single_evidence_control_d2_cd0aefac5e01", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "synthetic photorealistic scene of a yellow plastic toy car on a light blue fabric mat, clean composition, sharp focus", "target_evidence": "yellow plastic toy car", "expected_visual_elements": ["yellow plastic toy car", "light blue fabric mat", "clean composition"], "negative_prompt": "complex backgrounds, blurry focus, multiple objects", "diversity_tags": ["color", "material", "shape"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cd0aefac5e01", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4483, "diversity_score": 0.3517, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_cd0aefac5e01", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_cd0aefac5e01/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24641679, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_cd0aefac5e01/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_cd0aefac5e01/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_cd0aefac5e01/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5434002, "qwen_elapsed_seconds": 5.456, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_cd0aefac5e01/record.json"} +{"prompt_id": "single_evidence_control_d2_cd0cc398d97a", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "synthetic photorealistic scene of a white ceramic mug with a blue rim on a light wood surface, clean boundaries, high contrast", "target_evidence": "white ceramic mug", "expected_visual_elements": ["white ceramic mug", "blue rim", "light wood surface"], "negative_prompt": "distractor objects, blurry focus, complex background", "diversity_tags": ["color", "material", "surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cd0cc398d97a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_cd0cc398d97a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_cd0cc398d97a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28089663, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_cd0cc398d97a/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_cd0cc398d97a/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_cd0cc398d97a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6027182, "qwen_elapsed_seconds": 5.409, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_cd0cc398d97a/record.json"} +{"prompt_id": "single_evidence_control_d2_e101325bb6c3", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "synthetic photorealistic close up of a textured leather wallet with brown stitching on a white cloth background, high contrast, sharp edges", "target_evidence": "textured leather wallet", "expected_visual_elements": ["textured leather wallet", "brown stitching", "white cloth background"], "negative_prompt": "colorful patterns, soft focus, complex background", "diversity_tags": ["texture", "material", "color"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e101325bb6c3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.4231, "diversity_score": 0.3769, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_e101325bb6c3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_e101325bb6c3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23518460, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_e101325bb6c3/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_e101325bb6c3/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_e101325bb6c3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7394666, "qwen_elapsed_seconds": 5.996, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_e101325bb6c3/record.json"} +{"prompt_id": "single_evidence_control_d2_ef6eeadf6d27", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "Synthetic photorealistic scene of a single red rubber duck floating in clear water with sharp focus and minimal distractions.", "target_evidence": "red rubber duck", "expected_visual_elements": ["red rubber duck", "clear water"], "negative_prompt": "low resolution, blurry, distorted, cluttered background, extra objects", "diversity_tags": ["duck", "water", "floating"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ef6eeadf6d27", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4231, "diversity_score": 0.3769, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_ef6eeadf6d27", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_ef6eeadf6d27/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28844988, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_ef6eeadf6d27/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_ef6eeadf6d27/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_ef6eeadf6d27/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1498908, "qwen_elapsed_seconds": 5.844, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_ef6eeadf6d27/record.json"} +{"prompt_id": "single_evidence_control_d2_f1a05812badd", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "synthetic photorealistic image of a transparent blue glass sphere on a dark wooden table, clear foreground separation, high contrast", "target_evidence": "transparent blue glass sphere", "expected_visual_elements": ["transparent blue glass sphere", "dark wooden table", "clear boundaries"], "negative_prompt": "opaque surfaces, reflective textures, multiple objects", "diversity_tags": ["pattern", "transparency", "surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f1a05812badd", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.175, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_f1a05812badd", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_f1a05812badd/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28122596, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_f1a05812badd/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_f1a05812badd/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_f1a05812badd/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1740609, "qwen_elapsed_seconds": 5.865, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_f1a05812badd/record.json"} +{"prompt_id": "single_evidence_control_d2_f436c71b4cc9", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "Synthetic photorealistic view of a metallic silver cylinder standing on a dark wood table with clean lines and high contrast.", "target_evidence": "metallic silver cylinder", "expected_visual_elements": ["metallic silver cylinder", "dark wood table"], "negative_prompt": "low resolution, blurry, distorted, cluttered background, extra objects", "diversity_tags": ["cylinder", "metal", "table"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f436c71b4cc9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_f436c71b4cc9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_f436c71b4cc9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20612948, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_f436c71b4cc9/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_f436c71b4cc9/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_f436c71b4cc9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9617164, "qwen_elapsed_seconds": 6.074, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_f436c71b4cc9/record.json"} +{"prompt_id": "single_evidence_control_d2_f952e4bdb11b", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "Synthetic photorealistic close-up of a yellow rubber band stretched across a white plastic surface with high contrast and clean edges.", "target_evidence": "yellow rubber band", "expected_visual_elements": ["yellow rubber band", "white plastic surface"], "negative_prompt": "low resolution, blurry, distorted, cluttered background, extra objects", "diversity_tags": ["rubber band", "plastic", "stretched"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f952e4bdb11b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.6087, "diversity_score": 0.1913, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_f952e4bdb11b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_f952e4bdb11b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20634951, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_f952e4bdb11b/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_f952e4bdb11b/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_f952e4bdb11b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9824713, "qwen_elapsed_seconds": 6.123, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_f952e4bdb11b/record.json"} +{"prompt_id": "single_evidence_control_d2_fcb181c5b55f", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "synthetic photorealistic view of a glossy black cube with sharp edges on a matte gray platform, clean lines, high object-background contrast", "target_evidence": "glossy black cube", "expected_visual_elements": ["glossy black cube", "matte gray platform", "sharp edges"], "negative_prompt": "colorful elements, soft shadows, complex geometry", "diversity_tags": ["shape", "material", "geometry"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "fcb181c5b55f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4483, "diversity_score": 0.3517, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_fcb181c5b55f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_fcb181c5b55f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27505479, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_fcb181c5b55f/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_fcb181c5b55f/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_fcb181c5b55f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2064609, "qwen_elapsed_seconds": 6.425, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_fcb181c5b55f/record.json"} +{"prompt_id": "single_evidence_control_d2_fd0ba9f9c833", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 2, "image_prompt": "Synthetic photorealistic image of a green plastic cube placed on a textured concrete surface with sharp focus and clean edges.", "target_evidence": "green plastic cube", "expected_visual_elements": ["green plastic cube", "textured concrete surface"], "negative_prompt": "low resolution, blurry, distorted, cluttered background, extra objects", "diversity_tags": ["cube", "plastic", "concrete"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "fd0ba9f9c833", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d2_fd0ba9f9c833", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d2_fd0ba9f9c833/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27288820, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d2_fd0ba9f9c833/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_fd0ba9f9c833/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d2_fd0ba9f9c833/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 85512, "qwen_elapsed_seconds": 5.552, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d2_fd0ba9f9c833/record.json"} +{"prompt_id": "single_evidence_control_d3_21f3bcd8eb0f", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic wide-angle shot of a blue translucent glass cube floating on a dark matte background, clear foreground separation, sharp edges, and clean boundaries", "target_evidence": "blue translucent glass cube", "expected_visual_elements": ["blue translucent glass cube", "dark matte background"], "negative_prompt": "low resolution, blurry, extra objects, complex background, poor lighting", "diversity_tags": ["glass material", "floating object"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "21f3bcd8eb0f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wide angle shot", "max_similarity_to_other_prompt": 0.4828, "diversity_score": 0.3172, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_21f3bcd8eb0f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_21f3bcd8eb0f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20718310, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_21f3bcd8eb0f/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_21f3bcd8eb0f/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_21f3bcd8eb0f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1263112, "qwen_elapsed_seconds": 5.624, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_21f3bcd8eb0f/record.json"} +{"prompt_id": "single_evidence_control_d3_31cf9cbd130b", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic close-up of a rubber band stretched tightly on a smooth white surface with high contrast", "target_evidence": "rubber band", "expected_visual_elements": ["rubber band", "smooth white surface", "tight stretch"], "negative_prompt": "distractor objects, blurry edges, complex background", "diversity_tags": ["rubber", "stretched", "white surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "31cf9cbd130b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.6087, "diversity_score": 0.1913, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_31cf9cbd130b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_31cf9cbd130b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21593867, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_31cf9cbd130b/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_31cf9cbd130b/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_31cf9cbd130b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7570715, "qwen_elapsed_seconds": 5.625, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_31cf9cbd130b/record.json"} +{"prompt_id": "single_evidence_control_d3_5a037e0eb6c4", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic image of a smooth ceramic bowl with a glossy finish and high contrast against a plain gray background", "target_evidence": "ceramic bowl", "expected_visual_elements": ["ceramic bowl", "glossy finish", "gray background"], "negative_prompt": "distractor objects, low contrast, complex background", "diversity_tags": ["ceramic", "glossy", "gray background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5a037e0eb6c4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_5a037e0eb6c4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_5a037e0eb6c4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22790862, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_5a037e0eb6c4/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_5a037e0eb6c4/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_5a037e0eb6c4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6700471, "qwen_elapsed_seconds": 5.518, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_5a037e0eb6c4/record.json"} +{"prompt_id": "single_evidence_control_d3_632c72d86033", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic close-up of a smooth ceramic mug with glossy glaze, placed on a white ceramic plate with high contrast and sharp focus", "target_evidence": "glossy glaze on the mug", "expected_visual_elements": ["ceramic mug", "white ceramic plate", "glossy surface"], "negative_prompt": "blurry details, low contrast, background distractions, unrealistic lighting", "diversity_tags": ["ceramic", "glossy", "close-up"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "632c72d86033", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.5862, "diversity_score": 0.2138, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_632c72d86033", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_632c72d86033/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27309796, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_632c72d86033/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_632c72d86033/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_632c72d86033/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6021209, "qwen_elapsed_seconds": 5.546, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_632c72d86033/record.json"} +{"prompt_id": "single_evidence_control_d3_6d8401266afc", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic wide-angle shot of a wooden cube with sharp edges and clean lines on a dark wood floor", "target_evidence": "wooden cube", "expected_visual_elements": ["wooden cube", "dark wood floor", "clean lines"], "negative_prompt": "distractor objects, blurred edges, complex background", "diversity_tags": ["wooden", "cube", "wide-angle"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6d8401266afc", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wide angle shot", "max_similarity_to_other_prompt": 0.4828, "diversity_score": 0.3172, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_6d8401266afc", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_6d8401266afc/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23040153, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_6d8401266afc/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_6d8401266afc/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_6d8401266afc/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4474786, "qwen_elapsed_seconds": 5.557, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_6d8401266afc/record.json"} +{"prompt_id": "single_evidence_control_d3_72c15c1053ff", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic wide shot of a white ceramic tile with natural cracks, placed on a white ceramic floor with clean boundaries and high contrast", "target_evidence": "natural cracks on the tile", "expected_visual_elements": ["white ceramic tile", "white ceramic floor", "crack texture"], "negative_prompt": "blurry details, low contrast, background distractions, unrealistic lighting", "diversity_tags": ["ceramic", "cracks", "wide shot"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "72c15c1053ff", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wide shot of", "max_similarity_to_other_prompt": 0.5833, "diversity_score": 0.2167, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_72c15c1053ff", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_72c15c1053ff/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29214847, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_72c15c1053ff/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_72c15c1053ff/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_72c15c1053ff/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 273857, "qwen_elapsed_seconds": 5.521, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_72c15c1053ff/record.json"} +{"prompt_id": "single_evidence_control_d3_76efe04400e9", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic overhead view of a blue glass vial with clear surface, placed on a white glass surface with sharp focus and high contrast", "target_evidence": "clear surface of the vial", "expected_visual_elements": ["blue glass vial", "white glass surface", "clear texture"], "negative_prompt": "blurry details, low contrast, background distractions, unrealistic lighting", "diversity_tags": ["glass", "clear", "overhead"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "76efe04400e9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic overhead view of", "max_similarity_to_other_prompt": 0.6154, "diversity_score": 0.1846, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_76efe04400e9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_76efe04400e9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22174078, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_76efe04400e9/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_76efe04400e9/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_76efe04400e9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8336473, "qwen_elapsed_seconds": 5.527, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_76efe04400e9/record.json"} +{"prompt_id": "single_evidence_control_d3_9616afde47c3", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic close-up of a metal key with reflective surface, lying on a dark fabric with clear foreground separation and sharp focus", "target_evidence": "reflective surface of the key", "expected_visual_elements": ["metal key", "dark fabric", "reflective texture"], "negative_prompt": "blurry details, low contrast, background distractions, unrealistic lighting", "diversity_tags": ["metal", "reflective", "close-up"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "9616afde47c3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_9616afde47c3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_9616afde47c3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24076620, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_9616afde47c3/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_9616afde47c3/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_9616afde47c3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 703064, "qwen_elapsed_seconds": 5.484, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_9616afde47c3/record.json"} +{"prompt_id": "single_evidence_control_d3_9d6564a32416", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic close-up of a metal pipe with rusted surface, resting on a dark metal surface with clean boundaries and sharp focus", "target_evidence": "rusted surface of the pipe", "expected_visual_elements": ["metal pipe", "dark metal surface", "rusted texture"], "negative_prompt": "blurry details, low contrast, background distractions, unrealistic lighting", "diversity_tags": ["metal", "rusted", "close-up"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "9d6564a32416", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.5833, "diversity_score": 0.2167, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_9d6564a32416", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_9d6564a32416/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23162992, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_9d6564a32416/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_9d6564a32416/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_9d6564a32416/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2297016, "qwen_elapsed_seconds": 5.68, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_9d6564a32416/record.json"} +{"prompt_id": "single_evidence_control_d3_a3aee7af68a9", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic image of a smooth plastic plate with clear edges and high contrast on a plain light gray surface", "target_evidence": "plastic plate", "expected_visual_elements": ["plastic plate", "smooth surface", "light gray background"], "negative_prompt": "distractor objects, blurry edges, complex background", "diversity_tags": ["plastic", "plate", "light gray"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a3aee7af68a9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_a3aee7af68a9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_a3aee7af68a9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26212069, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_a3aee7af68a9/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_a3aee7af68a9/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_a3aee7af68a9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3071871, "qwen_elapsed_seconds": 5.616, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_a3aee7af68a9/record.json"} +{"prompt_id": "single_evidence_control_d3_b5a0a35c6155", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic view of a polished stone with sharp edges and clear boundaries on a light-colored stone floor", "target_evidence": "polished stone", "expected_visual_elements": ["polished stone", "sharp edges", "light-colored floor"], "negative_prompt": "distractor objects, low contrast, complex background", "diversity_tags": ["stone", "polished", "light-colored"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b5a0a35c6155", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4583, "diversity_score": 0.3417, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_b5a0a35c6155", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_b5a0a35c6155/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21107452, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_b5a0a35c6155/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_b5a0a35c6155/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_b5a0a35c6155/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 120502, "qwen_elapsed_seconds": 5.605, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_b5a0a35c6155/record.json"} +{"prompt_id": "single_evidence_control_d3_bc5d56105dab", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic close-up of a wooden box with carved patterns, resting on a wooden table with clean boundaries and sharp focus", "target_evidence": "carved patterns on the box", "expected_visual_elements": ["wooden box", "wooden table", "carved design"], "negative_prompt": "blurry details, low contrast, background distractions, unrealistic lighting", "diversity_tags": ["carved", "wooden", "close-up"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "bc5d56105dab", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.5833, "diversity_score": 0.2167, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_bc5d56105dab", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_bc5d56105dab/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28562766, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_bc5d56105dab/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_bc5d56105dab/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_bc5d56105dab/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 716134, "qwen_elapsed_seconds": 5.187, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_bc5d56105dab/record.json"} +{"prompt_id": "single_evidence_control_d3_bfd6ba140016", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic close-up of a rough wooden sphere on a smooth white surface, high contrast lighting, clean boundaries, sharp focus, and clear foreground separation", "target_evidence": "rough wooden sphere", "expected_visual_elements": ["rough wooden sphere", "smooth white surface"], "negative_prompt": "low resolution, blurry, extra objects, complex background, poor lighting", "diversity_tags": ["wooden texture", "high contrast lighting"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "bfd6ba140016", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.7407, "diversity_score": 0.0593, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_bfd6ba140016", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_bfd6ba140016/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29450634, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_bfd6ba140016/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_bfd6ba140016/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_bfd6ba140016/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8759630, "qwen_elapsed_seconds": 6.158, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_bfd6ba140016/record.json"} +{"prompt_id": "single_evidence_control_d3_c2898a7bc0a6", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic overhead view of a smooth glass tile with reflections and sharp focus on its surface texture", "target_evidence": "glass tile surface texture", "expected_visual_elements": ["glass tile", "reflections", "surface texture"], "negative_prompt": "distorted reflections, blurred surface, complex background", "diversity_tags": ["glass", "overhead", "texture"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c2898a7bc0a6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic overhead view of", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_c2898a7bc0a6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_c2898a7bc0a6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26285167, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_c2898a7bc0a6/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_c2898a7bc0a6/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_c2898a7bc0a6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1969252, "qwen_elapsed_seconds": 5.362, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_c2898a7bc0a6/record.json"} +{"prompt_id": "single_evidence_control_d3_c9c20e24a039", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic overhead view of a black rubber band with smooth surface, placed on a black rubber mat with sharp focus and high contrast", "target_evidence": "smooth surface of the rubber band", "expected_visual_elements": ["black rubber band", "black rubber mat", "smooth texture"], "negative_prompt": "blurry details, low contrast, background distractions, unrealistic lighting", "diversity_tags": ["rubber", "smooth", "overhead"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c9c20e24a039", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic overhead view of", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.175, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_c9c20e24a039", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_c9c20e24a039/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20902976, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_c9c20e24a039/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_c9c20e24a039/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_c9c20e24a039/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8184165, "qwen_elapsed_seconds": 5.269, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_c9c20e24a039/record.json"} +{"prompt_id": "single_evidence_control_d3_cf7397980b78", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic wide shot of a green plastic bottle with smooth surface, sitting on a green plastic tray with clean boundaries and high contrast", "target_evidence": "smooth surface of the bottle", "expected_visual_elements": ["green plastic bottle", "green plastic tray", "smooth texture"], "negative_prompt": "blurry details, low contrast, background distractions, unrealistic lighting", "diversity_tags": ["plastic", "smooth", "wide shot"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cf7397980b78", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wide shot of", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_cf7397980b78", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_cf7397980b78/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29745822, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_cf7397980b78/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_cf7397980b78/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_cf7397980b78/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 263039, "qwen_elapsed_seconds": 5.47, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_cf7397980b78/record.json"} +{"prompt_id": "single_evidence_control_d3_df625edf8c51", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic overhead view of a red rubber ball on a black rubber mat with sharp focus and clean surface separation", "target_evidence": "red color of the ball", "expected_visual_elements": ["red rubber ball", "black rubber mat", "high contrast"], "negative_prompt": "blurry details, low contrast, background distractions, unrealistic lighting", "diversity_tags": ["color", "rubber", "overhead"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "df625edf8c51", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic overhead view of", "max_similarity_to_other_prompt": 0.6957, "diversity_score": 0.1043, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_df625edf8c51", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_df625edf8c51/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23372186, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_df625edf8c51/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_df625edf8c51/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_df625edf8c51/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5924800, "qwen_elapsed_seconds": 5.98, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_df625edf8c51/record.json"} +{"prompt_id": "single_evidence_control_d3_e279f213497b", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic wide shot of a wooden spoon with natural grain texture, resting on a wooden cutting board with clean boundaries and high object-background contrast", "target_evidence": "natural grain texture of the spoon", "expected_visual_elements": ["wooden spoon", "wooden cutting board", "grain texture"], "negative_prompt": "blurry details, low contrast, background distractions, unrealistic lighting", "diversity_tags": ["wooden", "texture", "wide shot"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e279f213497b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wide shot of", "max_similarity_to_other_prompt": 0.4828, "diversity_score": 0.3172, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_e279f213497b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_e279f213497b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29473205, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_e279f213497b/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_e279f213497b/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_e279f213497b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 218887, "qwen_elapsed_seconds": 6.433, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_e279f213497b/record.json"} +{"prompt_id": "single_evidence_control_d3_f5b138e143c1", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic overhead shot of a reflective mirror with clean edges and high contrast on a dark background", "target_evidence": "reflective mirror", "expected_visual_elements": ["reflective mirror", "clean edges", "dark background"], "negative_prompt": "distractor objects, low contrast, complex background", "diversity_tags": ["mirror", "reflective", "overhead"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f5b138e143c1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic overhead shot of", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_f5b138e143c1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_f5b138e143c1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 30108950, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_f5b138e143c1/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_f5b138e143c1/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_f5b138e143c1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2783541, "qwen_elapsed_seconds": 6.661, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_f5b138e143c1/record.json"} +{"prompt_id": "single_evidence_control_d3_fdf74144016f", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 3, "image_prompt": "synthetic photorealistic close-up of a metallic sphere on a matte black surface with high contrast and clean boundaries", "target_evidence": "metallic sphere", "expected_visual_elements": ["metallic sphere", "matte black surface"], "negative_prompt": "distractor objects, low contrast, blurry edges, complex background", "diversity_tags": ["metallic", "matte black", "close-up"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "fdf74144016f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d3_fdf74144016f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d3_fdf74144016f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27353136, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d3_fdf74144016f/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_fdf74144016f/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d3_fdf74144016f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2841104, "qwen_elapsed_seconds": 5.358, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d3_fdf74144016f/record.json"} +{"prompt_id": "single_evidence_control_d4_05538498dfe3", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic scene of a red rubber ball on a green grassy surface with sharp focus and clean edges", "target_evidence": "red rubber ball", "expected_visual_elements": ["red rubber ball", "green grassy surface"], "negative_prompt": "blurry, low contrast, multiple objects, cluttered background", "diversity_tags": ["rubber", "grass", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "05538498dfe3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.6522, "diversity_score": 0.1478, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_05538498dfe3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_05538498dfe3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24105386, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_05538498dfe3/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_05538498dfe3/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_05538498dfe3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5742199, "qwen_elapsed_seconds": 5.53, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_05538498dfe3/record.json"} +{"prompt_id": "single_evidence_control_d4_05ce097a3bc9", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic shot of a smooth ceramic mug with a glossy finish, sitting on a white marble countertop, high contrast between object and background, clean boundaries, sharp focus", "target_evidence": "glossy ceramic mug", "expected_visual_elements": ["ceramic mug", "white marble countertop"], "negative_prompt": "blurry, low contrast, cluttered background, multiple objects, shadows, reflections", "diversity_tags": ["ceramic", "marble", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "05ce097a3bc9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.5333, "diversity_score": 0.2667, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_05ce097a3bc9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_05ce097a3bc9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26073792, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_05ce097a3bc9/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_05ce097a3bc9/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_05ce097a3bc9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3024228, "qwen_elapsed_seconds": 5.471, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_05ce097a3bc9/record.json"} +{"prompt_id": "single_evidence_control_d4_240f9cbb10f5", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic image of a metallic silver sphere resting on a dark gray concrete floor, sharp focus, clear boundaries, high contrast between object and background", "target_evidence": "metallic silver sphere", "expected_visual_elements": ["silver sphere", "concrete floor"], "negative_prompt": "low resolution, blurred edges, multiple objects, reflections", "diversity_tags": ["metallic", "concrete", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "240f9cbb10f5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.4516, "diversity_score": 0.3484, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_240f9cbb10f5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_240f9cbb10f5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23752216, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_240f9cbb10f5/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_240f9cbb10f5/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_240f9cbb10f5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9590679, "qwen_elapsed_seconds": 5.099, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_240f9cbb10f5/record.json"} +{"prompt_id": "single_evidence_control_d4_433dda18796a", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic shot of a green plastic container on a white tile floor with sharp focus and clear edges", "target_evidence": "green plastic container", "expected_visual_elements": ["green plastic container", "white tile floor"], "negative_prompt": "low resolution, poor lighting, multiple objects, cluttered background", "diversity_tags": ["plastic", "tile", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "433dda18796a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_433dda18796a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_433dda18796a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22532850, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_433dda18796a/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_433dda18796a/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_433dda18796a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 862556, "qwen_elapsed_seconds": 5.273, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_433dda18796a/record.json"} +{"prompt_id": "single_evidence_control_d4_4eb150435f04", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic image of a wooden crate with a carved wooden pattern on its surface, placed on a smooth tiled floor, clean background, sharp focus", "target_evidence": "carved wooden pattern", "expected_visual_elements": ["wooden crate", "tiled floor"], "negative_prompt": "blurry, low resolution, complex background, overlapping objects", "diversity_tags": ["wood", "carved", "tiled"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4eb150435f04", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.4286, "diversity_score": 0.3714, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_4eb150435f04", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_4eb150435f04/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29768040, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_4eb150435f04/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_4eb150435f04/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_4eb150435f04/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2710941, "qwen_elapsed_seconds": 5.205, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_4eb150435f04/record.json"} +{"prompt_id": "single_evidence_control_d4_516e6a0c91c1", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic shot of a smooth white ceramic mug with a red stripe, sitting on a textured concrete surface under soft overhead lighting, clean boundaries and high contrast", "target_evidence": "red stripe on ceramic mug", "expected_visual_elements": ["white ceramic mug", "red stripe", "textured concrete surface", "soft overhead lighting"], "negative_prompt": "low resolution, blurry, extra objects, complex background, poor lighting", "diversity_tags": ["ceramic", "stripe", "concrete", "overhead lighting"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "516e6a0c91c1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.5333, "diversity_score": 0.2667, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_516e6a0c91c1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_516e6a0c91c1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28689386, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_516e6a0c91c1/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_516e6a0c91c1/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_516e6a0c91c1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3449745, "qwen_elapsed_seconds": 5.235, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_516e6a0c91c1/record.json"} +{"prompt_id": "single_evidence_control_d4_686253c3a450", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic close up of a polished silver spoon with a carved handle, on a white marble countertop, under bright studio lighting, with clear boundaries and high contrast", "target_evidence": "carved handle on silver spoon", "expected_visual_elements": ["polished silver spoon", "carved handle", "white marble countertop", "bright studio lighting"], "negative_prompt": "low resolution, blurry, extra objects, complex background, poor lighting", "diversity_tags": ["metal", "handle", "marble", "studio lighting"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "686253c3a450", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.4062, "diversity_score": 0.3937, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_686253c3a450", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_686253c3a450/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22023812, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_686253c3a450/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_686253c3a450/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_686253c3a450/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4931026, "qwen_elapsed_seconds": 5.845, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_686253c3a450/record.json"} +{"prompt_id": "single_evidence_control_d4_694fb75d5368", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic wide shot of a smooth black rubber ball on a white tile floor, under soft ambient lighting, with clean boundaries and high contrast", "target_evidence": "smooth black rubber ball", "expected_visual_elements": ["smooth black rubber ball", "white tile floor", "soft ambient lighting", "clean boundaries"], "negative_prompt": "low resolution, blurry, extra objects, complex background, poor lighting", "diversity_tags": ["rubber", "ball", "tile", "ambient lighting"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "694fb75d5368", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wide shot of", "max_similarity_to_other_prompt": 0.5714, "diversity_score": 0.2286, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_694fb75d5368", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_694fb75d5368/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23273275, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_694fb75d5368/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_694fb75d5368/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_694fb75d5368/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7357520, "qwen_elapsed_seconds": 6.714, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_694fb75d5368/record.json"} +{"prompt_id": "single_evidence_control_d4_768263068ab3", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a white foam sponge on a dark blue fabric surface, clean boundaries, sharp focus, clear foreground separation, high contrast", "target_evidence": "white foam sponge", "expected_visual_elements": ["foam sponge", "blue fabric"], "negative_prompt": "distorted texture, low resolution, overlapping objects, shadows", "diversity_tags": ["foam", "fabric", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "768263068ab3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5517, "diversity_score": 0.2483, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_768263068ab3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_768263068ab3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21148794, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_768263068ab3/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_768263068ab3/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_768263068ab3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5190593, "qwen_elapsed_seconds": 6.732, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_768263068ab3/record.json"} +{"prompt_id": "single_evidence_control_d4_7f3df0cacb8f", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a transparent red glass cube on a dark blue surface, with even lighting and sharp focus, clean boundaries and high contrast", "target_evidence": "transparent red glass cube", "expected_visual_elements": ["transparent red glass cube", "dark blue surface", "even lighting", "sharp focus"], "negative_prompt": "low resolution, blurry, extra objects, complex background, poor lighting", "diversity_tags": ["glass", "cube", "blue", "even lighting"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7f3df0cacb8f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5357, "diversity_score": 0.2643, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_7f3df0cacb8f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_7f3df0cacb8f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24367900, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_7f3df0cacb8f/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_7f3df0cacb8f/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_7f3df0cacb8f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9778951, "qwen_elapsed_seconds": 6.725, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_7f3df0cacb8f/record.json"} +{"prompt_id": "single_evidence_control_d4_86dd66b3e456", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a geometrically patterned fabric draped over a wooden chair, clean surface, high object-background contrast, clear foreground separation", "target_evidence": "geometric pattern on fabric", "expected_visual_elements": ["patterned fabric", "wooden chair"], "negative_prompt": "messy texture, overlapping objects, low resolution, complex background", "diversity_tags": ["fabric", "wooden", "pattern"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "86dd66b3e456", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_86dd66b3e456", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_86dd66b3e456/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23591552, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_86dd66b3e456/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_86dd66b3e456/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_86dd66b3e456/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4910570, "qwen_elapsed_seconds": 6.83, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_86dd66b3e456/record.json"} +{"prompt_id": "single_evidence_control_d4_af4c62bf826b", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic overhead shot of a green plastic container with a raised rim, on a light brown wooden surface, with even lighting and clear separation", "target_evidence": "raised rim on green plastic container", "expected_visual_elements": ["green plastic container", "raised rim", "light brown wooden surface", "even lighting"], "negative_prompt": "low resolution, blurry, extra objects, complex background, poor lighting", "diversity_tags": ["plastic", "container", "wooden", "even lighting"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "af4c62bf826b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic overhead shot of", "max_similarity_to_other_prompt": 0.4286, "diversity_score": 0.3714, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_af4c62bf826b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_af4c62bf826b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22556406, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_af4c62bf826b/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_af4c62bf826b/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_af4c62bf826b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8515615, "qwen_elapsed_seconds": 6.851, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_af4c62bf826b/record.json"} +{"prompt_id": "single_evidence_control_d4_be8f0dc59246", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a red rubber ball on a light beige wall, high contrast, clean background, sharp focus, clear foreground separation", "target_evidence": "red rubber ball", "expected_visual_elements": ["rubber ball", "beige wall"], "negative_prompt": "low contrast, cluttered background, multiple objects, shadows", "diversity_tags": ["rubber", "wall", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "be8f0dc59246", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4828, "diversity_score": 0.3172, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_be8f0dc59246", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_be8f0dc59246/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20377066, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_be8f0dc59246/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_be8f0dc59246/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_be8f0dc59246/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3366961, "qwen_elapsed_seconds": 5.467, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_be8f0dc59246/record.json"} +{"prompt_id": "single_evidence_control_d4_bf4102b5d83f", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic shot of a black leather wallet on a light-colored wooden desk, clear boundaries, high contrast, sharp focus, clean background", "target_evidence": "black leather wallet", "expected_visual_elements": ["leather wallet", "wooden desk"], "negative_prompt": "low contrast, cluttered background, multiple objects, shadows", "diversity_tags": ["leather", "wooden", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "bf4102b5d83f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.5185, "diversity_score": 0.2815, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_bf4102b5d83f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_bf4102b5d83f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28135887, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_bf4102b5d83f/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_bf4102b5d83f/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_bf4102b5d83f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5239783, "qwen_elapsed_seconds": 5.521, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_bf4102b5d83f/record.json"} +{"prompt_id": "single_evidence_control_d4_c53054c30f1a", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic shot of a translucent glass bowl filled with water, sitting on a textured stone surface, clean boundaries, sharp focus, clear foreground separation", "target_evidence": "translucent glass bowl", "expected_visual_elements": ["glass bowl", "textured stone"], "negative_prompt": "distorted reflections, cluttered background, low resolution, shadows", "diversity_tags": ["glass", "stone", "translucent"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c53054c30f1a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.4375, "diversity_score": 0.3625, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_c53054c30f1a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_c53054c30f1a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29387446, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_c53054c30f1a/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_c53054c30f1a/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_c53054c30f1a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3301617, "qwen_elapsed_seconds": 5.478, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_c53054c30f1a/record.json"} +{"prompt_id": "single_evidence_control_d4_dcd8e4ad5dfc", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic image of a translucent blue glass vase with a geometric pattern, placed on a dark wooden table under directional lighting, with high contrast and clean separation", "target_evidence": "geometric pattern on glass vase", "expected_visual_elements": ["translucent blue glass vase", "geometric pattern", "dark wooden table", "directional lighting"], "negative_prompt": "low resolution, blurry, extra objects, complex background, poor lighting", "diversity_tags": ["glass", "pattern", "wooden", "directional lighting"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "dcd8e4ad5dfc", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.4828, "diversity_score": 0.3172, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_dcd8e4ad5dfc", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_dcd8e4ad5dfc/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21211921, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_dcd8e4ad5dfc/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_dcd8e4ad5dfc/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_dcd8e4ad5dfc/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3492674, "qwen_elapsed_seconds": 5.421, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_dcd8e4ad5dfc/record.json"} +{"prompt_id": "single_evidence_control_d4_de585bf843a1", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic view of a matte black metal sphere resting on a pale green fabric, with even lighting and sharp focus, clear foreground separation and clean edges", "target_evidence": "matte black metal sphere", "expected_visual_elements": ["matte black metal sphere", "pale green fabric", "even lighting", "sharp focus"], "negative_prompt": "low resolution, blurry, extra objects, complex background, poor lighting", "diversity_tags": ["metal", "sphere", "fabric", "even lighting"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "de585bf843a1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4848, "diversity_score": 0.3152, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_de585bf843a1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_de585bf843a1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21563092, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_de585bf843a1/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_de585bf843a1/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_de585bf843a1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9241276, "qwen_elapsed_seconds": 5.349, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_de585bf843a1/record.json"} +{"prompt_id": "single_evidence_control_d4_df96457bab87", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic image of a white plastic chair with a curved backrest, on a dark green carpet, under soft lighting, with clean separation and high contrast", "target_evidence": "curved backrest on white plastic chair", "expected_visual_elements": ["white plastic chair", "curved backrest", "dark green carpet", "soft lighting"], "negative_prompt": "low resolution, blurry, extra objects, complex background, poor lighting", "diversity_tags": ["plastic", "chair", "carpet", "soft lighting"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "df96457bab87", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.4545, "diversity_score": 0.3455, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_df96457bab87", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_df96457bab87/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29287907, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_df96457bab87/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_df96457bab87/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_df96457bab87/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8945774, "qwen_elapsed_seconds": 5.375, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_df96457bab87/record.json"} +{"prompt_id": "single_evidence_control_d4_ff000cba7e79", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic scene of a yellow plastic toy car on a dark grey floor, with strong lighting and sharp focus, clean foreground separation and high contrast", "target_evidence": "yellow plastic toy car", "expected_visual_elements": ["yellow plastic toy car", "dark grey floor", "strong lighting", "sharp focus"], "negative_prompt": "low resolution, blurry, extra objects, complex background, poor lighting", "diversity_tags": ["plastic", "toy", "grey", "strong lighting"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ff000cba7e79", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.4483, "diversity_score": 0.3517, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_ff000cba7e79", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_ff000cba7e79/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23543658, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_ff000cba7e79/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_ff000cba7e79/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_ff000cba7e79/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1934909, "qwen_elapsed_seconds": 5.477, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_ff000cba7e79/record.json"} +{"prompt_id": "single_evidence_control_d4_ffcf882a2cff", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 4, "image_prompt": "synthetic photorealistic shot of a silver metallic cylinder with a patterned surface, on a light blue fabric, under directional lighting, with clear foreground separation", "target_evidence": "patterned surface on metallic cylinder", "expected_visual_elements": ["silver metallic cylinder", "patterned surface", "light blue fabric", "directional lighting"], "negative_prompt": "low resolution, blurry, extra objects, complex background, poor lighting", "diversity_tags": ["metallic", "cylinder", "fabric", "directional lighting"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ffcf882a2cff", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.4, "diversity_score": 0.4, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d4_ffcf882a2cff", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d4_ffcf882a2cff/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27892953, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d4_ffcf882a2cff/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_ffcf882a2cff/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d4_ffcf882a2cff/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3605137, "qwen_elapsed_seconds": 5.449, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d4_ffcf882a2cff/record.json"} +{"prompt_id": "single_evidence_control_d5_0f0933d7b485", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic wide shot of a red rubber ball on a green felt, sharp focus, high contrast, isolated object in a clean scene", "target_evidence": "one central target object", "expected_visual_elements": ["red rubber ball", "green felt", "clean background"], "negative_prompt": "blurry edges, low contrast, complex environment, multiple objects", "diversity_tags": ["ball", "felt", "green"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0f0933d7b485", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wide shot of", "max_similarity_to_other_prompt": 0.5185, "diversity_score": 0.2815, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_0f0933d7b485", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_0f0933d7b485/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24575918, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_0f0933d7b485/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_0f0933d7b485/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_0f0933d7b485/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3542284, "qwen_elapsed_seconds": 5.597, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_0f0933d7b485/record.json"} +{"prompt_id": "single_evidence_control_d5_17dbc0e1bae3", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic image of a single black rubber duck with a smooth surface, placed on a white textured surface, clear separation, and high contrast", "target_evidence": "single black rubber duck", "expected_visual_elements": ["single black rubber duck", "white textured surface"], "negative_prompt": "multiple ducks, low contrast, cluttered surface, shadows", "diversity_tags": ["rubber material", "textured surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "17dbc0e1bae3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5185, "diversity_score": 0.2815, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_17dbc0e1bae3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_17dbc0e1bae3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26038989, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_17dbc0e1bae3/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_17dbc0e1bae3/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_17dbc0e1bae3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2588545, "qwen_elapsed_seconds": 5.326, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_17dbc0e1bae3/record.json"} +{"prompt_id": "single_evidence_control_d5_2148bb0faf37", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic close up of a translucent glass cube with a blue gradient, placed on a dark fabric, clean separation, sharp focus, isolated object", "target_evidence": "one central target object", "expected_visual_elements": ["translucent glass cube", "blue gradient", "dark fabric"], "negative_prompt": "low resolution, color distortion, complex patterns, multiple objects", "diversity_tags": ["glass", "gradient", "dark fabric"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2148bb0faf37", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.4333, "diversity_score": 0.3667, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_2148bb0faf37", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_2148bb0faf37/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21032507, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_2148bb0faf37/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_2148bb0faf37/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_2148bb0faf37/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3557488, "qwen_elapsed_seconds": 5.388, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_2148bb0faf37/record.json"} +{"prompt_id": "single_evidence_control_d5_26dc9efdcd8a", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic overhead view of a wooden plank with a natural grain pattern, placed on a light gray surface, high contrast, clean background", "target_evidence": "one obvious attribute such as pattern", "expected_visual_elements": ["wooden plank", "natural grain", "light gray surface"], "negative_prompt": "low detail, color imbalance, complex textures, multiple objects", "diversity_tags": ["wood", "grain", "light gray"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "26dc9efdcd8a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic overhead view of", "max_similarity_to_other_prompt": 0.4643, "diversity_score": 0.3357, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_26dc9efdcd8a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_26dc9efdcd8a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24446114, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_26dc9efdcd8a/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_26dc9efdcd8a/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_26dc9efdcd8a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6363494, "qwen_elapsed_seconds": 5.415, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_26dc9efdcd8a/record.json"} +{"prompt_id": "single_evidence_control_d5_43abc2775b67", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic close-up of a smooth ceramic mug with a subtle crack pattern, against a plain gray background, sharp focus, and clean separation", "target_evidence": "smooth ceramic mug", "expected_visual_elements": ["smooth ceramic mug", "plain gray background"], "negative_prompt": "blurry mug, multiple cracks, complex background, reflective surfaces", "diversity_tags": ["ceramic surface", "subtle pattern"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "43abc2775b67", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.4333, "diversity_score": 0.3667, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_43abc2775b67", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_43abc2775b67/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29214842, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_43abc2775b67/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_43abc2775b67/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_43abc2775b67/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5827298, "qwen_elapsed_seconds": 5.309, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_43abc2775b67/record.json"} +{"prompt_id": "single_evidence_control_d5_454014145150", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic shot of a translucent glass sphere with intricate internal patterns, surrounded by a plain white backdrop, sharp focus, and clear separation", "target_evidence": "translucent glass sphere", "expected_visual_elements": ["translucent glass sphere", "plain white backdrop"], "negative_prompt": "distorted patterns, reflective surfaces, multiple objects, complex background", "diversity_tags": ["glass material", "internal patterns"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "454014145150", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.3871, "diversity_score": 0.4129, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_454014145150", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_454014145150/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29701681, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_454014145150/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_454014145150/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_454014145150/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8769424, "qwen_elapsed_seconds": 5.384, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_454014145150/record.json"} +{"prompt_id": "single_evidence_control_d5_48d6abb580b9", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic close up of a green leaf with a translucent vein, placed on a light green background, high contrast, clean boundaries", "target_evidence": "one central target object", "expected_visual_elements": ["green leaf", "translucent vein", "light green background"], "negative_prompt": "blurry edges, low contrast, complex textures, multiple objects", "diversity_tags": ["leaf", "vein", "green"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "48d6abb580b9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic close up of", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_48d6abb580b9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_48d6abb580b9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22868265, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_48d6abb580b9/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_48d6abb580b9/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_48d6abb580b9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9918257, "qwen_elapsed_seconds": 5.319, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_48d6abb580b9/record.json"} +{"prompt_id": "single_evidence_control_d5_4c5484153bb8", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic shot of a single metal key with a unique engraving pattern, placed on a dark wooden surface, clear foreground, and high contrast", "target_evidence": "single metal key", "expected_visual_elements": ["single metal key", "dark wooden surface"], "negative_prompt": "multiple keys, low contrast, cluttered surface, shadows", "diversity_tags": ["engraved metal", "wooden surface"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4c5484153bb8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.5556, "diversity_score": 0.2444, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_4c5484153bb8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_4c5484153bb8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29780903, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_4c5484153bb8/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_4c5484153bb8/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_4c5484153bb8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7719862, "qwen_elapsed_seconds": 5.346, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_4c5484153bb8/record.json"} +{"prompt_id": "single_evidence_control_d5_52c925338512", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic shot of a metallic sphere with a matte texture on a white background, clean boundaries, high contrast, sharp focus, isolated object", "target_evidence": "one obvious attribute such as material", "expected_visual_elements": ["metallic sphere", "matte texture", "white background"], "negative_prompt": "low contrast, blurred edges, complex background, multiple objects", "diversity_tags": ["material", "sphere", "white background"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "52c925338512", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.5185, "diversity_score": 0.2815, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_52c925338512", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_52c925338512/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27195612, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_52c925338512/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_52c925338512/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_52c925338512/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6765139, "qwen_elapsed_seconds": 5.314, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_52c925338512/record.json"} +{"prompt_id": "single_evidence_control_d5_6742c237235e", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a single white ceramic tile with a subtle blue glaze, on a plain black background, clear foreground separation, and high contrast", "target_evidence": "single white ceramic tile", "expected_visual_elements": ["single white ceramic tile", "plain black background"], "negative_prompt": "multiple tiles, low contrast, complex background, shadows", "diversity_tags": ["ceramic glaze", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6742c237235e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4828, "diversity_score": 0.3172, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_6742c237235e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_6742c237235e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25843954, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_6742c237235e/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_6742c237235e/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_6742c237235e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9700867, "qwen_elapsed_seconds": 5.337, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_6742c237235e/record.json"} +{"prompt_id": "single_evidence_control_d5_6be4aae78a54", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a single red rubber band stretched across a white table, clean boundaries, high contrast, and clear foreground focus", "target_evidence": "red rubber band", "expected_visual_elements": ["red rubber band", "white table"], "negative_prompt": "multiple rubber bands, low contrast, cluttered table, shadows", "diversity_tags": ["elastic material", "color contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6be4aae78a54", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.4643, "diversity_score": 0.3357, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_6be4aae78a54", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_6be4aae78a54/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24369635, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_6be4aae78a54/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_6be4aae78a54/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_6be4aae78a54/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9968334, "qwen_elapsed_seconds": 5.379, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_6be4aae78a54/record.json"} +{"prompt_id": "single_evidence_control_d5_71d95745fb73", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic image of a polished brass gear with high contrast against a matte black background, clean boundaries, sharp focus, and clear foreground separation", "target_evidence": "polished brass gear", "expected_visual_elements": ["polished brass gear", "matte black background"], "negative_prompt": "low contrast, blurry edges, multiple objects, cluttered background", "diversity_tags": ["metallic surface", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "71d95745fb73", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_71d95745fb73", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_71d95745fb73/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24469217, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_71d95745fb73/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_71d95745fb73/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_71d95745fb73/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9918517, "qwen_elapsed_seconds": 5.363, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_71d95745fb73/record.json"} +{"prompt_id": "single_evidence_control_d5_94aa2339354f", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic overhead shot of a white ceramic mug with a cracked surface, placed on a dark wooden surface, clean separation, high contrast", "target_evidence": "one obvious attribute such as material", "expected_visual_elements": ["white ceramic mug", "cracked surface", "dark wooden surface"], "negative_prompt": "low detail, color imbalance, complex background, multiple objects", "diversity_tags": ["ceramic", "crack", "wood"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "94aa2339354f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic overhead shot of", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_94aa2339354f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_94aa2339354f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28980430, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_94aa2339354f/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_94aa2339354f/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_94aa2339354f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3814227, "qwen_elapsed_seconds": 5.373, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_94aa2339354f/record.json"} +{"prompt_id": "single_evidence_control_d5_98389805a08b", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic view of a textured ceramic tile with a subtle crack, placed on a white ceramic tile, clean boundaries, high contrast", "target_evidence": "one obvious attribute such as pattern", "expected_visual_elements": ["textured ceramic tile", "subtle crack", "white ceramic tile"], "negative_prompt": "low resolution, color distortion, complex background, multiple objects", "diversity_tags": ["ceramic", "crack", "white"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "98389805a08b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic view of a", "max_similarity_to_other_prompt": 0.5833, "diversity_score": 0.2167, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_98389805a08b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_98389805a08b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25617434, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_98389805a08b/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_98389805a08b/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_98389805a08b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3631960, "qwen_elapsed_seconds": 5.371, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_98389805a08b/record.json"} +{"prompt_id": "single_evidence_control_d5_acac8b09899a", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic wide angle shot of a black rubber eraser with a smooth surface, placed on a light blue background, sharp focus, high contrast", "target_evidence": "one central target object", "expected_visual_elements": ["black rubber eraser", "smooth surface", "light blue background"], "negative_prompt": "low resolution, color distortion, complex environment, multiple objects", "diversity_tags": ["rubber", "smooth", "blue"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "acac8b09899a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wide angle shot", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_acac8b09899a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_acac8b09899a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21994801, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_acac8b09899a/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_acac8b09899a/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_acac8b09899a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6436926, "qwen_elapsed_seconds": 5.41, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_acac8b09899a/record.json"} +{"prompt_id": "single_evidence_control_d5_b27bea471d92", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic shot of a single red plastic cube with a glossy finish, on a white non-reflective surface, clear foreground, and high contrast", "target_evidence": "single red plastic cube", "expected_visual_elements": ["single red plastic cube", "white non-reflective surface"], "negative_prompt": "multiple cubes, low contrast, cluttered surface, shadows", "diversity_tags": ["plastic material", "glossy finish"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b27bea471d92", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.4667, "diversity_score": 0.3333, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_b27bea471d92", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_b27bea471d92/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27374824, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_b27bea471d92/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_b27bea471d92/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_b27bea471d92/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 647570, "qwen_elapsed_seconds": 5.436, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_b27bea471d92/record.json"} +{"prompt_id": "single_evidence_control_d5_bfb7feb04712", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic image of a single leaf with vein patterns on a light green background, clear foreground separation, and high contrast", "target_evidence": "single leaf", "expected_visual_elements": ["single leaf", "light green background"], "negative_prompt": "multiple leaves, low contrast, cluttered background, shadows", "diversity_tags": ["natural surface", "vein patterns"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "bfb7feb04712", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_bfb7feb04712", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_bfb7feb04712/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24043108, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_bfb7feb04712/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_bfb7feb04712/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_bfb7feb04712/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1538947, "qwen_elapsed_seconds": 5.215, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_bfb7feb04712/record.json"} +{"prompt_id": "single_evidence_control_d5_cd15d13311f1", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic scene of a textured linen fabric draped over a smooth wooden frame, with clear foreground separation and high surface contrast", "target_evidence": "textured linen fabric", "expected_visual_elements": ["textured linen fabric", "smooth wooden frame"], "negative_prompt": "blurry fabric, low contrast, multiple objects, reflective surfaces", "diversity_tags": ["textured surface", "wooden frame"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cd15d13311f1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic scene of a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_cd15d13311f1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_cd15d13311f1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28692851, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_cd15d13311f1/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_cd15d13311f1/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_cd15d13311f1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4540614, "qwen_elapsed_seconds": 5.251, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_cd15d13311f1/record.json"} +{"prompt_id": "single_evidence_control_d5_ea199a81a704", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic shot of a blue plastic bowl with a smooth surface, sitting on a dark wooden table, high contrast, clean separation", "target_evidence": "one central target object", "expected_visual_elements": ["blue plastic bowl", "smooth surface", "dark wooden table"], "negative_prompt": "low detail, color imbalance, complex textures, multiple objects", "diversity_tags": ["plastic", "smooth", "wood"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ea199a81a704", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic shot of a", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_ea199a81a704", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_ea199a81a704/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29745422, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_ea199a81a704/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_ea199a81a704/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_ea199a81a704/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 810765, "qwen_elapsed_seconds": 5.283, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_ea199a81a704/record.json"} +{"prompt_id": "single_evidence_control_d5_ed3f660899f1", "run_name": "flux2_qil_decomposition_pilot", "category": "single_evidence_control", "difficulty": 5, "image_prompt": "synthetic photorealistic image of a yellow fabric with a wavy pattern on a black background, sharp focus, high contrast, isolated object", "target_evidence": "one obvious attribute such as pattern", "expected_visual_elements": ["yellow fabric", "wavy pattern", "black background"], "negative_prompt": "low resolution, color distortion, complex environment, multiple objects", "diversity_tags": ["fabric", "pattern", "black"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ed3f660899f1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic image of a", "max_similarity_to_other_prompt": 0.4815, "diversity_score": 0.3185, "repeated_opening": true, "review": {"prompt_id": "single_evidence_control_d5_ed3f660899f1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 5, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 1.0, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/single_evidence_control/single_evidence_control_d5_ed3f660899f1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27774177, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/single_evidence_control/single_evidence_control_d5_ed3f660899f1/layers/layer_00.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_ed3f660899f1/layers/layer_01.png", "layers-x/k3/single_evidence_control/single_evidence_control_d5_ed3f660899f1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5539321, "qwen_elapsed_seconds": 5.214, "qwen_record_path": "layers-x/k3/single_evidence_control/single_evidence_control_d5_ed3f660899f1/record.json"} +{"prompt_id": "visual_knowledge_d1_05f4d81ad8bb", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic equipment label with a flame warning icon on a metal surface, sharp focus and high contrast for easy decomposition.", "target_evidence": "flame warning icon", "expected_visual_elements": ["equipment label", "flame warning icon", "metal surface"], "negative_prompt": "low detail, soft focus, curved edges, watermarks", "diversity_tags": ["equipment label", "fire safety", "sharp focus"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "05f4d81ad8bb", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment label with", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_05f4d81ad8bb", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_05f4d81ad8bb/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22288887, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_05f4d81ad8bb/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_05f4d81ad8bb/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_05f4d81ad8bb/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5800170, "qwen_elapsed_seconds": 5.205, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_05f4d81ad8bb/record.json"} +{"prompt_id": "visual_knowledge_d1_09a5fe350e7d", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic panel with a fork and knife icon on a dark blue panel, clean and separated for easy decomposition.", "target_evidence": "fork and knife icon", "expected_visual_elements": ["panel", "fork and knife icon", "dark blue background"], "negative_prompt": "overly detailed, low resolution, curved edges, watermarks", "diversity_tags": ["panel", "food service", "clean edges"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "09a5fe350e7d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic panel with a", "max_similarity_to_other_prompt": 0.6087, "diversity_score": 0.1913, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_09a5fe350e7d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_09a5fe350e7d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27390473, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_09a5fe350e7d/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_09a5fe350e7d/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_09a5fe350e7d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 260884, "qwen_elapsed_seconds": 5.293, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_09a5fe350e7d/record.json"} +{"prompt_id": "visual_knowledge_d1_18172281824d", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic dashboard panel showing a red medical cross icon against a dark gray panel with clean edges for clear symbol isolation.", "target_evidence": "red medical cross", "expected_visual_elements": ["dashboard panel", "red medical cross", "dark gray background"], "negative_prompt": "overly detailed, low resolution, reflections, shadows", "diversity_tags": ["dashboard", "medical icon", "clean edges"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "18172281824d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard panel showing", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_18172281824d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_18172281824d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21686386, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_18172281824d/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_18172281824d/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_18172281824d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3337348, "qwen_elapsed_seconds": 5.223, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_18172281824d/record.json"} +{"prompt_id": "visual_knowledge_d1_1d17a5dab778", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic equipment label with a fork and knife icon on a light yellow background, isolated and clean for decomposition.", "target_evidence": "fork and knife icon", "expected_visual_elements": ["equipment label", "fork and knife icon"], "negative_prompt": "low contrast, mixed colors, overlapping text, grainy texture", "diversity_tags": ["equipment label", "food service"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1d17a5dab778", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment label with", "max_similarity_to_other_prompt": 0.5909, "diversity_score": 0.2091, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_1d17a5dab778", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_1d17a5dab778/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22898057, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_1d17a5dab778/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_1d17a5dab778/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_1d17a5dab778/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2070083, "qwen_elapsed_seconds": 5.206, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_1d17a5dab778/record.json"} +{"prompt_id": "visual_knowledge_d1_205a630fec90", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic wall sticker with a wheelchair accessibility icon on a light blue wall, clear boundaries for symbol separation.", "target_evidence": "wheelchair accessibility icon", "expected_visual_elements": ["wall sticker", "wheelchair accessibility icon", "light blue wall"], "negative_prompt": "distorted, low resolution, transparent background, shadows", "diversity_tags": ["wall sticker", "accessibility", "clear boundaries"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "205a630fec90", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wall sticker with", "max_similarity_to_other_prompt": 0.619, "diversity_score": 0.181, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_205a630fec90", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_205a630fec90/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27357525, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_205a630fec90/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_205a630fec90/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_205a630fec90/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8772868, "qwen_elapsed_seconds": 5.242, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_205a630fec90/record.json"} +{"prompt_id": "visual_knowledge_d1_2085eed6f031", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic wall sticker showing a yellow lightning bolt on a dark blue wall, sharply focused and distinct from surroundings.", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["wall sticker", "yellow lightning bolt"], "negative_prompt": "overexposed, washed out colors, faded sticker, cluttered wall", "diversity_tags": ["wall sticker", "electrical hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "2085eed6f031", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wall sticker showing", "max_similarity_to_other_prompt": 0.4231, "diversity_score": 0.3769, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_2085eed6f031", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_2085eed6f031/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24665057, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_2085eed6f031/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_2085eed6f031/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_2085eed6f031/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6962100, "qwen_elapsed_seconds": 5.165, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_2085eed6f031/record.json"} +{"prompt_id": "visual_knowledge_d1_30ad0cff5bb8", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic dashboard panel with a skull warning icon on a dark grey panel, isolated and sharp for clear symbol identification.", "target_evidence": "skull warning icon", "expected_visual_elements": ["dashboard panel", "skull warning icon"], "negative_prompt": "distracting background, low detail, faded icon, blurry edges", "diversity_tags": ["dashboard panel", "toxic hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "30ad0cff5bb8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard panel with", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_30ad0cff5bb8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_30ad0cff5bb8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21487495, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_30ad0cff5bb8/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_30ad0cff5bb8/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_30ad0cff5bb8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2175661, "qwen_elapsed_seconds": 5.274, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_30ad0cff5bb8/record.json"} +{"prompt_id": "visual_knowledge_d1_394fb0f7b3c9", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic folder with a water droplet icon on a white background, sharply focused and separated from folder text.", "target_evidence": "water droplet icon", "expected_visual_elements": ["folder", "water droplet icon"], "negative_prompt": "faded icon, low contrast, complex folder design, cluttered background", "diversity_tags": ["folder", "water source"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "394fb0f7b3c9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic folder with a", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_394fb0f7b3c9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_394fb0f7b3c9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23689293, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_394fb0f7b3c9/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_394fb0f7b3c9/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_394fb0f7b3c9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5247764, "qwen_elapsed_seconds": 5.218, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_394fb0f7b3c9/record.json"} +{"prompt_id": "visual_knowledge_d1_3ff10b51b3fa", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic box with a green recycling icon on a light green box, clean and sharp for easy decomposition.", "target_evidence": "green recycling icon", "expected_visual_elements": ["box", "green recycling icon", "light green color"], "negative_prompt": "blurry, low contrast, cartoon style, watermarks", "diversity_tags": ["box", "recycling", "clean design"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "3ff10b51b3fa", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic box with a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_3ff10b51b3fa", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_3ff10b51b3fa/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20452128, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_3ff10b51b3fa/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_3ff10b51b3fa/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_3ff10b51b3fa/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5932000, "qwen_elapsed_seconds": 5.175, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_3ff10b51b3fa/record.json"} +{"prompt_id": "visual_knowledge_d1_71e2e71fb91f", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "synthetic photorealistic control panel with a yellow lightning bolt icon on a dark blue background, clean lines, high contrast, sharp focus", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["control panel", "dark blue background"], "negative_prompt": "blurry, low contrast, overlapping elements, complex background", "diversity_tags": ["control panel", "lightning bolt"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "71e2e71fb91f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic control panel with", "max_similarity_to_other_prompt": 0.6957, "diversity_score": 0.1043, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_71e2e71fb91f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_71e2e71fb91f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20657490, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_71e2e71fb91f/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_71e2e71fb91f/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_71e2e71fb91f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9110965, "qwen_elapsed_seconds": 5.132, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_71e2e71fb91f/record.json"} +{"prompt_id": "visual_knowledge_d1_7865459ca825", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic notice board with a yellow lightning bolt warning sign on a clean white background, high contrast for easy decomposition.", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["notice board", "yellow lightning bolt", "white background"], "negative_prompt": "blurry, low contrast, cartoon style, watermarks", "diversity_tags": ["notice", "warning sign", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7865459ca825", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notice board with", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.175, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_7865459ca825", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_7865459ca825/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28122230, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_7865459ca825/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_7865459ca825/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_7865459ca825/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7792157, "qwen_elapsed_seconds": 5.234, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_7865459ca825/record.json"} +{"prompt_id": "visual_knowledge_d1_7865b7495838", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic equipment tag with a green recycling icon on a yellow background, high contrast and clearly separated from the tag text.", "target_evidence": "green recycling icon", "expected_visual_elements": ["equipment tag", "green recycling icon"], "negative_prompt": "low resolution, mixed colors, overlapping text, grainy texture", "diversity_tags": ["equipment tag", "recycling"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7865b7495838", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment tag with", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_7865b7495838", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_7865b7495838/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23797773, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_7865b7495838/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_7865b7495838/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_7865b7495838/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4109197, "qwen_elapsed_seconds": 5.15, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_7865b7495838/record.json"} +{"prompt_id": "visual_knowledge_d1_7a433eee9be2", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic control panel with a red medical cross symbol on a white background, clean and isolated for clear decomposition.", "target_evidence": "red medical cross", "expected_visual_elements": ["control panel", "red medical cross"], "negative_prompt": "blurry, low contrast, complex background, overlapping elements", "diversity_tags": ["control panel", "medical aid"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7a433eee9be2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic control panel with", "max_similarity_to_other_prompt": 0.7083, "diversity_score": 0.0917, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_7a433eee9be2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_7a433eee9be2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20757595, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_7a433eee9be2/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_7a433eee9be2/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_7a433eee9be2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9935787, "qwen_elapsed_seconds": 5.208, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_7a433eee9be2/record.json"} +{"prompt_id": "visual_knowledge_d1_be2c43905015", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "synthetic photorealistic equipment tag with a yellow lightning bolt on a metal surface, clean boundaries, high contrast, sharp focus", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["equipment tag", "yellow lightning bolt"], "negative_prompt": "blurry, low contrast, background distractions, poor lighting", "diversity_tags": ["equipment tag", "electrical hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "be2c43905015", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment tag with", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_be2c43905015", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_be2c43905015/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21894526, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_be2c43905015/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_be2c43905015/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_be2c43905015/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9913126, "qwen_elapsed_seconds": 5.268, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_be2c43905015/record.json"} +{"prompt_id": "visual_knowledge_d1_c900c67e85ef", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic dashboard with a leaf eco icon on a light green dashboard, high contrast for symbol isolation.", "target_evidence": "leaf eco icon", "expected_visual_elements": ["dashboard", "leaf eco icon", "light green color"], "negative_prompt": "blurry, low contrast, cartoon style, watermarks", "diversity_tags": ["dashboard", "eco-friendly", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c900c67e85ef", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard with a", "max_similarity_to_other_prompt": 0.5238, "diversity_score": 0.2762, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_c900c67e85ef", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_c900c67e85ef/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21111854, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_c900c67e85ef/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_c900c67e85ef/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_c900c67e85ef/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4337891, "qwen_elapsed_seconds": 5.223, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_c900c67e85ef/record.json"} +{"prompt_id": "visual_knowledge_d1_ce74eafd04a3", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic dashboard with a wifi icon on a light grey panel, high contrast and clearly separated from other indicators.", "target_evidence": "wifi icon", "expected_visual_elements": ["dashboard", "wifi icon"], "negative_prompt": "overlapping elements, low detail, poor lighting, grainy texture", "diversity_tags": ["dashboard", "wireless access"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ce74eafd04a3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard with a", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_ce74eafd04a3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_ce74eafd04a3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29437989, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_ce74eafd04a3/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_ce74eafd04a3/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_ce74eafd04a3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7139251, "qwen_elapsed_seconds": 5.277, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_ce74eafd04a3/record.json"} +{"prompt_id": "visual_knowledge_d1_d2fe6d6049ed", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic folder with a water droplet icon on a light yellow folder, high contrast for symbol visibility.", "target_evidence": "water droplet icon", "expected_visual_elements": ["folder", "water droplet icon", "light yellow color"], "negative_prompt": "low resolution, soft focus, reflections, watermarks", "diversity_tags": ["folder", "water source", "high contrast"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d2fe6d6049ed", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic folder with a", "max_similarity_to_other_prompt": 0.5238, "diversity_score": 0.2762, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_d2fe6d6049ed", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_d2fe6d6049ed/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26513479, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_d2fe6d6049ed/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_d2fe6d6049ed/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_d2fe6d6049ed/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 435526, "qwen_elapsed_seconds": 5.191, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_d2fe6d6049ed/record.json"} +{"prompt_id": "visual_knowledge_d1_dc571287a462", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic box with a leaf eco icon on a light green background, clean and sharply defined for easy decomposition.", "target_evidence": "leaf eco icon", "expected_visual_elements": ["box", "leaf eco icon"], "negative_prompt": "overlapping elements, low contrast, poor lighting, grainy texture", "diversity_tags": ["box", "eco-friendly"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "dc571287a462", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic box with a", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_dc571287a462", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_dc571287a462/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28801591, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_dc571287a462/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_dc571287a462/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_dc571287a462/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1041454, "qwen_elapsed_seconds": 5.211, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_dc571287a462/record.json"} +{"prompt_id": "visual_knowledge_d1_f1e64aceee5e", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic notice board with a wheelchair accessibility icon on a light blue background, clear and distinct from other symbols.", "target_evidence": "wheelchair accessibility icon", "expected_visual_elements": ["notice board", "wheelchair accessibility icon"], "negative_prompt": "overexposed, low resolution, faded icon, cluttered board", "diversity_tags": ["notice board", "accessibility"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f1e64aceee5e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notice board with", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_f1e64aceee5e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_f1e64aceee5e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26827328, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_f1e64aceee5e/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_f1e64aceee5e/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_f1e64aceee5e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8626609, "qwen_elapsed_seconds": 5.22, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_f1e64aceee5e/record.json"} +{"prompt_id": "visual_knowledge_d1_f8ab59cd5ff8", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 1, "image_prompt": "Synthetic photorealistic equipment tag with a wifi icon on a light gray tag, clean and isolated for easy decomposition.", "target_evidence": "wifi icon", "expected_visual_elements": ["equipment tag", "wifi icon", "light gray color"], "negative_prompt": "overly detailed, low resolution, curved edges, watermarks", "diversity_tags": ["equipment tag", "wireless access", "clean design"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f8ab59cd5ff8", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment tag with", "max_similarity_to_other_prompt": 0.5909, "diversity_score": 0.2091, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d1_f8ab59cd5ff8", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d1_f8ab59cd5ff8/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29521790, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d1_f8ab59cd5ff8/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_f8ab59cd5ff8/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d1_f8ab59cd5ff8/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9908869, "qwen_elapsed_seconds": 5.261, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d1_f8ab59cd5ff8/record.json"} +{"prompt_id": "visual_knowledge_d2_0b8ceb2874b0", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic equipment tag showing a yellow lightning bolt against a dark metal surface, clear boundaries and high lighting for symbol isolation", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["equipment tag", "yellow lightning bolt"], "negative_prompt": "overexposed highlights, overlapping elements, soft focus", "diversity_tags": ["equipment tag", "electrical hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0b8ceb2874b0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment tag showing", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_0b8ceb2874b0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_0b8ceb2874b0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21972515, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_0b8ceb2874b0/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_0b8ceb2874b0/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_0b8ceb2874b0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7893507, "qwen_elapsed_seconds": 5.158, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_0b8ceb2874b0/record.json"} +{"prompt_id": "visual_knowledge_d2_1cac844d136a", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic wall sticker with a red medical cross on a light gray wall, clean edges and flat lighting for easy symbol separation", "target_evidence": "red medical cross", "expected_visual_elements": ["wall sticker", "red medical cross"], "negative_prompt": "textured wall, shadows, complex background", "diversity_tags": ["wall sticker", "medical aid"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1cac844d136a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wall sticker with", "max_similarity_to_other_prompt": 0.4444, "diversity_score": 0.3556, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_1cac844d136a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_1cac844d136a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29341089, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_1cac844d136a/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_1cac844d136a/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_1cac844d136a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9232544, "qwen_elapsed_seconds": 5.157, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_1cac844d136a/record.json"} +{"prompt_id": "visual_knowledge_d2_29e6503e1672", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic control panel with a green recycling icon on a clean white background, high contrast and sharp focus for easy decomposition", "target_evidence": "green recycling icon", "expected_visual_elements": ["control panel", "green recycling icon"], "negative_prompt": "blurry, low contrast, complex background, overlapping elements", "diversity_tags": ["control panel", "recycling"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "29e6503e1672", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic control panel with", "max_similarity_to_other_prompt": 0.75, "diversity_score": 0.05, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_29e6503e1672", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_29e6503e1672/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25625688, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_29e6503e1672/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_29e6503e1672/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_29e6503e1672/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2267821, "qwen_elapsed_seconds": 5.189, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_29e6503e1672/record.json"} +{"prompt_id": "visual_knowledge_d2_381678b652c4", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic panel with a skull warning icon on a clean white surface, strong lighting and clear boundaries", "target_evidence": "skull warning icon", "expected_visual_elements": ["panel", "skull warning icon"], "negative_prompt": "blurry, low contrast, complex background, text", "diversity_tags": ["panel", "toxic hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "381678b652c4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic panel with a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_381678b652c4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_381678b652c4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24071540, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_381678b652c4/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_381678b652c4/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_381678b652c4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2665877, "qwen_elapsed_seconds": 5.197, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_381678b652c4/record.json"} +{"prompt_id": "visual_knowledge_d2_3cef4fd299ef", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic equipment label with a leaf eco icon on a metal tag, clean lines and bright lighting for easy decomposition", "target_evidence": "leaf eco icon", "expected_visual_elements": ["equipment label", "leaf eco icon"], "negative_prompt": "faded, low resolution, complex background, text", "diversity_tags": ["equipment label", "eco-friendly"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "3cef4fd299ef", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment label with", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_3cef4fd299ef", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_3cef4fd299ef/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24277945, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_3cef4fd299ef/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_3cef4fd299ef/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_3cef4fd299ef/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1495998, "qwen_elapsed_seconds": 5.221, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_3cef4fd299ef/record.json"} +{"prompt_id": "visual_knowledge_d2_65340d6eb5d3", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic panel with a fork and knife icon on a dark blue background, clean edges and sharp lighting for symbol isolation", "target_evidence": "fork and knife icon", "expected_visual_elements": ["panel", "fork and knife icon"], "negative_prompt": "overlapping elements, low contrast, soft focus", "diversity_tags": ["panel", "food service"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "65340d6eb5d3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic panel with a", "max_similarity_to_other_prompt": 0.6087, "diversity_score": 0.1913, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_65340d6eb5d3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_65340d6eb5d3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28502821, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_65340d6eb5d3/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_65340d6eb5d3/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_65340d6eb5d3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1085383, "qwen_elapsed_seconds": 5.225, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_65340d6eb5d3/record.json"} +{"prompt_id": "visual_knowledge_d2_7106059e8a9b", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic dashboard with a wifi icon on a dark panel, high contrast and sharp focus for clear visibility", "target_evidence": "wifi icon", "expected_visual_elements": ["dashboard", "wifi icon"], "negative_prompt": "low contrast, cluttered background, overlapping elements, text", "diversity_tags": ["dashboard", "wireless access"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7106059e8a9b", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard with a", "max_similarity_to_other_prompt": 0.5417, "diversity_score": 0.2583, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_7106059e8a9b", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_7106059e8a9b/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24727357, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_7106059e8a9b/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_7106059e8a9b/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_7106059e8a9b/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1966110, "qwen_elapsed_seconds": 5.183, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_7106059e8a9b/record.json"} +{"prompt_id": "visual_knowledge_d2_7585179e2b66", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic dashboard panel showing a wifi icon on a dark gray background, clean boundaries and high lighting for symbol separation", "target_evidence": "wifi icon", "expected_visual_elements": ["dashboard panel", "wifi icon"], "negative_prompt": "distracting elements, soft focus, low contrast", "diversity_tags": ["dashboard panel", "wireless access"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7585179e2b66", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard panel showing", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_7585179e2b66", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_7585179e2b66/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25271771, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_7585179e2b66/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_7585179e2b66/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_7585179e2b66/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6095574, "qwen_elapsed_seconds": 5.26, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_7585179e2b66/record.json"} +{"prompt_id": "visual_knowledge_d2_7c8be66b7ec2", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic folder with a water droplet icon on its cover, clean background and high contrast for clear decomposition", "target_evidence": "water droplet icon", "expected_visual_elements": ["folder", "water droplet icon"], "negative_prompt": "faded, low resolution, complex background, text", "diversity_tags": ["folder", "water source"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7c8be66b7ec2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic folder with a", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.175, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_7c8be66b7ec2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_7c8be66b7ec2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23924477, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_7c8be66b7ec2/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_7c8be66b7ec2/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_7c8be66b7ec2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8515820, "qwen_elapsed_seconds": 5.227, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_7c8be66b7ec2/record.json"} +{"prompt_id": "visual_knowledge_d2_805b659daa38", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic equipment tag with a red medical cross on a metal surface, clean lines and bright lighting for clear visibility", "target_evidence": "red medical cross", "expected_visual_elements": ["equipment tag", "red medical cross"], "negative_prompt": "faded, low resolution, background distractions, text overlay", "diversity_tags": ["equipment tag", "medical aid"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "805b659daa38", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment tag with", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_805b659daa38", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_805b659daa38/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27231427, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_805b659daa38/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_805b659daa38/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_805b659daa38/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5534981, "qwen_elapsed_seconds": 5.221, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_805b659daa38/record.json"} +{"prompt_id": "visual_knowledge_d2_85adab82c114", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic folder with a water droplet icon on a transparent surface, high contrast and clear foreground separation for easy decomposition", "target_evidence": "water droplet icon", "expected_visual_elements": ["folder", "water droplet icon"], "negative_prompt": "faded colors, background distractions, low detail", "diversity_tags": ["folder", "water source"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "85adab82c114", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic folder with a", "max_similarity_to_other_prompt": 0.625, "diversity_score": 0.175, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_85adab82c114", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_85adab82c114/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20835375, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_85adab82c114/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_85adab82c114/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_85adab82c114/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 839520, "qwen_elapsed_seconds": 5.239, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_85adab82c114/record.json"} +{"prompt_id": "visual_knowledge_d2_aae27359989a", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic notice board with a fork and knife icon on a wooden surface, sharp focus and high contrast", "target_evidence": "fork and knife icon", "expected_visual_elements": ["notice board", "fork and knife icon"], "negative_prompt": "low contrast, cluttered background, overlapping elements, text", "diversity_tags": ["notice board", "food service"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "aae27359989a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notice board with", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_aae27359989a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_aae27359989a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24024913, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_aae27359989a/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_aae27359989a/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_aae27359989a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6352024, "qwen_elapsed_seconds": 5.231, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_aae27359989a/record.json"} +{"prompt_id": "visual_knowledge_d2_c377df847c7d", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic dashboard panel showing a flame warning icon against a dark gray panel, high contrast and sharp focus", "target_evidence": "flame warning icon", "expected_visual_elements": ["dashboard panel", "flame warning icon"], "negative_prompt": "low contrast, cluttered background, overlapping elements, text", "diversity_tags": ["dashboard panel", "fire safety"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c377df847c7d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard panel showing", "max_similarity_to_other_prompt": 0.5217, "diversity_score": 0.2783, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_c377df847c7d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_c377df847c7d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26993853, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_c377df847c7d/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_c377df847c7d/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_c377df847c7d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5437797, "qwen_elapsed_seconds": 5.322, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_c377df847c7d/record.json"} +{"prompt_id": "visual_knowledge_d2_cd6fac3234a0", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic box with a wheelchair accessibility icon on a light blue surface, sharp focus and clean lines for clear symbol identification", "target_evidence": "wheelchair accessibility icon", "expected_visual_elements": ["box", "wheelchair accessibility icon"], "negative_prompt": "complex patterns, soft focus, cluttered background", "diversity_tags": ["box", "accessibility"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "cd6fac3234a0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic box with a", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_cd6fac3234a0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_cd6fac3234a0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29827040, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_cd6fac3234a0/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_cd6fac3234a0/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_cd6fac3234a0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5496906, "qwen_elapsed_seconds": 5.216, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_cd6fac3234a0/record.json"} +{"prompt_id": "visual_knowledge_d2_d3de109dbf41", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic control panel with a skull warning icon on a light green background, sharp focus and clear foreground for easy decomposition", "target_evidence": "skull warning icon", "expected_visual_elements": ["control panel", "skull warning icon"], "negative_prompt": "overexposed highlights, complex patterns, low detail", "diversity_tags": ["control panel", "toxic hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d3de109dbf41", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic control panel with", "max_similarity_to_other_prompt": 0.6154, "diversity_score": 0.1846, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_d3de109dbf41", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_d3de109dbf41/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28275334, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_d3de109dbf41/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_d3de109dbf41/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_d3de109dbf41/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9961061, "qwen_elapsed_seconds": 5.215, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_d3de109dbf41/record.json"} +{"prompt_id": "visual_knowledge_d2_da81900c70b1", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic dashboard panel showing a red medical cross on a dark blue surface, clear foreground separation, and simple lighting for precise symbol isolation", "target_evidence": "red medical cross", "expected_visual_elements": ["red medical cross", "dark blue surface", "dashboard panel"], "negative_prompt": "low resolution, distracting elements, reflections, shadows, complex textures", "diversity_tags": ["dashboard panel", "medical aid"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "da81900c70b1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard panel showing", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_da81900c70b1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_da81900c70b1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22188837, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_da81900c70b1/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_da81900c70b1/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_da81900c70b1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3905507, "qwen_elapsed_seconds": 5.22, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_da81900c70b1/record.json"} +{"prompt_id": "visual_knowledge_d2_e095332e1c4e", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic equipment label with a leaf eco icon on a yellow surface, high contrast and flat lighting for clear symbol visibility", "target_evidence": "leaf eco icon", "expected_visual_elements": ["equipment label", "leaf eco icon"], "negative_prompt": "textured surface, shadows, low resolution", "diversity_tags": ["equipment label", "eco-friendly"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e095332e1c4e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment label with", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_e095332e1c4e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_e095332e1c4e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27618430, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_e095332e1c4e/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_e095332e1c4e/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_e095332e1c4e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1110173, "qwen_elapsed_seconds": 5.204, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_e095332e1c4e/record.json"} +{"prompt_id": "visual_knowledge_d2_e5d7deaf583f", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic wall sticker with a yellow lightning bolt symbol on a plain beige wall, clear boundaries and strong lighting", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["wall sticker", "yellow lightning bolt"], "negative_prompt": "overexposed, out of focus, text overlay, complex pattern", "diversity_tags": ["wall sticker", "electrical hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e5d7deaf583f", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wall sticker with", "max_similarity_to_other_prompt": 0.4286, "diversity_score": 0.3714, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_e5d7deaf583f", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_e5d7deaf583f/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28508702, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_e5d7deaf583f/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_e5d7deaf583f/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_e5d7deaf583f/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8878777, "qwen_elapsed_seconds": 5.227, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_e5d7deaf583f/record.json"} +{"prompt_id": "visual_knowledge_d2_fc22de4b6f8c", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 2, "image_prompt": "synthetic photorealistic wall sticker with a yellow lightning bolt on a white background, clean boundaries, sharp focus, and high contrast for easy decomposition", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["yellow lightning bolt", "white background", "wall sticker"], "negative_prompt": "blurry, low contrast, complex background, overlapping objects, watermarks", "diversity_tags": ["wall sticker", "electrical hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "fc22de4b6f8c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wall sticker with", "max_similarity_to_other_prompt": 0.6154, "diversity_score": 0.1846, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d2_fc22de4b6f8c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d2_fc22de4b6f8c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28850957, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d2_fc22de4b6f8c/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_fc22de4b6f8c/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d2_fc22de4b6f8c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1744048, "qwen_elapsed_seconds": 5.262, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d2_fc22de4b6f8c/record.json"} +{"prompt_id": "visual_knowledge_d3_0120f18a72b6", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic dashboard showing a red medical cross icon on a dark blue panel, clean boundaries, clear foreground separation", "target_evidence": "red medical cross", "expected_visual_elements": ["dashboard", "red medical cross"], "negative_prompt": "distorted, low resolution, background clutter, text labels", "diversity_tags": ["dashboard", "medical icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0120f18a72b6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard showing a", "max_similarity_to_other_prompt": 0.5769, "diversity_score": 0.2231, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_0120f18a72b6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_0120f18a72b6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24230515, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_0120f18a72b6/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_0120f18a72b6/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_0120f18a72b6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7985871, "qwen_elapsed_seconds": 5.229, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_0120f18a72b6/record.json"} +{"prompt_id": "visual_knowledge_d3_23c087fbd8ac", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic wall sticker with a red medical cross icon, high contrast against a neutral gray wall, clear foreground separation", "target_evidence": "red medical cross", "expected_visual_elements": ["wall sticker", "medical icon"], "negative_prompt": "faded colors, unclear boundaries, cluttered background, low resolution", "diversity_tags": ["wall sticker", "medical icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "23c087fbd8ac", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wall sticker with", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_23c087fbd8ac", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_23c087fbd8ac/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23607424, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_23c087fbd8ac/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_23c087fbd8ac/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_23c087fbd8ac/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5505344, "qwen_elapsed_seconds": 5.206, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_23c087fbd8ac/record.json"} +{"prompt_id": "visual_knowledge_d3_39a2962afc8a", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic equipment label showing a green recycling icon on a dark blue panel with high contrast edges", "target_evidence": "green recycling icon", "expected_visual_elements": ["equipment label", "green recycling icon", "dark blue panel"], "negative_prompt": "low contrast, background noise, unclear boundaries", "diversity_tags": ["equipment label", "recycling"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "39a2962afc8a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment label showing", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.5, "repeated_opening": false, "review": {"prompt_id": "visual_knowledge_d3_39a2962afc8a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_39a2962afc8a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26806605, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_39a2962afc8a/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_39a2962afc8a/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_39a2962afc8a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6738476, "qwen_elapsed_seconds": 5.237, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_39a2962afc8a/record.json"} +{"prompt_id": "visual_knowledge_d3_406c29ed07d7", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic notice board with a red medical cross symbol, surrounded by clean text labels and clear boundaries", "target_evidence": "red medical cross", "expected_visual_elements": ["notice board", "red medical cross", "text labels"], "negative_prompt": "distorted, low resolution, overlapping text, poor lighting", "diversity_tags": ["notice board", "medical aid"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "406c29ed07d7", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notice board with", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_406c29ed07d7", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_406c29ed07d7/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23223050, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_406c29ed07d7/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_406c29ed07d7/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_406c29ed07d7/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6516705, "qwen_elapsed_seconds": 5.199, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_406c29ed07d7/record.json"} +{"prompt_id": "visual_knowledge_d3_60ba2f5c2b30", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic notice board with a water droplet icon on a wooden surface, clean background and sharp focus on the symbol", "target_evidence": "water droplet icon", "expected_visual_elements": ["notice board", "icon"], "negative_prompt": "faded colors, poor lighting, unclear boundaries, low contrast", "diversity_tags": ["notice board", "icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "60ba2f5c2b30", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notice board with", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_60ba2f5c2b30", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_60ba2f5c2b30/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29556540, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_60ba2f5c2b30/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_60ba2f5c2b30/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_60ba2f5c2b30/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4670917, "qwen_elapsed_seconds": 5.253, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_60ba2f5c2b30/record.json"} +{"prompt_id": "visual_knowledge_d3_8ad0881c5b3d", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic dashboard panel with a fork and knife icon on a dark gray background, clean edges and clear foreground separation", "target_evidence": "fork and knife icon", "expected_visual_elements": ["dashboard panel", "food service icon"], "negative_prompt": "overlapping elements, low contrast, blurry symbol, cluttered background", "diversity_tags": ["dashboard panel", "food service icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "8ad0881c5b3d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard panel with", "max_similarity_to_other_prompt": 0.56, "diversity_score": 0.24, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_8ad0881c5b3d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_8ad0881c5b3d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28255216, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_8ad0881c5b3d/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_8ad0881c5b3d/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_8ad0881c5b3d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1527705, "qwen_elapsed_seconds": 5.232, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_8ad0881c5b3d/record.json"} +{"prompt_id": "visual_knowledge_d3_a6ea0241fbc1", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic equipment tag with a skull warning icon on a metal surface, high contrast and clear symbol placement", "target_evidence": "skull warning icon", "expected_visual_elements": ["equipment tag", "warning icon"], "negative_prompt": "low resolution, poor lighting, blurry symbol, cluttered background", "diversity_tags": ["equipment tag", "warning icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a6ea0241fbc1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment tag with", "max_similarity_to_other_prompt": 0.75, "diversity_score": 0.05, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_a6ea0241fbc1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_a6ea0241fbc1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24155202, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_a6ea0241fbc1/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_a6ea0241fbc1/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_a6ea0241fbc1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1282101, "qwen_elapsed_seconds": 5.215, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_a6ea0241fbc1/record.json"} +{"prompt_id": "visual_knowledge_d3_aa3480b35ce1", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic wall sticker showing a red medical cross on a white surface, high contrast and clearly visible", "target_evidence": "red medical cross", "expected_visual_elements": ["wall sticker", "red medical cross"], "negative_prompt": "low resolution, faded colors, background distractions", "diversity_tags": ["wall sticker", "medical aid"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "aa3480b35ce1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wall sticker showing", "max_similarity_to_other_prompt": 0.4583, "diversity_score": 0.3417, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_aa3480b35ce1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_aa3480b35ce1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20578616, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_aa3480b35ce1/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_aa3480b35ce1/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_aa3480b35ce1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6150850, "qwen_elapsed_seconds": 5.248, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_aa3480b35ce1/record.json"} +{"prompt_id": "visual_knowledge_d3_ae38e3abca31", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic wall sticker with a leaf eco icon, surrounded by a green border, high contrast, sharp focus", "target_evidence": "leaf eco icon", "expected_visual_elements": ["wall sticker", "leaf eco icon"], "negative_prompt": "low detail, faded colors, background distractions, watermarks", "diversity_tags": ["wall sticker", "eco icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ae38e3abca31", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wall sticker with", "max_similarity_to_other_prompt": 0.44, "diversity_score": 0.36, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_ae38e3abca31", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_ae38e3abca31/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23905208, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_ae38e3abca31/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_ae38e3abca31/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_ae38e3abca31/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2531271, "qwen_elapsed_seconds": 5.221, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_ae38e3abca31/record.json"} +{"prompt_id": "visual_knowledge_d3_b569c616ec50", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic box with a water droplet icon on a light gray surface, high contrast and clear focus", "target_evidence": "water droplet icon", "expected_visual_elements": ["box", "water droplet icon", "light gray surface"], "negative_prompt": "blurry, low contrast, background noise", "diversity_tags": ["box", "water source"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b569c616ec50", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic box with a", "max_similarity_to_other_prompt": 0.619, "diversity_score": 0.181, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_b569c616ec50", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_b569c616ec50/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21278069, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_b569c616ec50/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_b569c616ec50/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_b569c616ec50/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 135996, "qwen_elapsed_seconds": 5.181, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_b569c616ec50/record.json"} +{"prompt_id": "visual_knowledge_d3_c3d4454f9721", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic equipment label with a flame warning icon, placed on a red surface, clear foreground, sharp focus", "target_evidence": "flame warning icon", "expected_visual_elements": ["equipment label", "flame warning icon"], "negative_prompt": "faded, low resolution, text labels, background noise", "diversity_tags": ["equipment label", "fire icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c3d4454f9721", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment label with", "max_similarity_to_other_prompt": 0.5455, "diversity_score": 0.2545, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_c3d4454f9721", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_c3d4454f9721/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27374760, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_c3d4454f9721/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_c3d4454f9721/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_c3d4454f9721/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4753514, "qwen_elapsed_seconds": 5.287, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_c3d4454f9721/record.json"} +{"prompt_id": "visual_knowledge_d3_d1f3728e53c1", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic dashboard panel with a green recycling icon on a light beige background, high contrast, clear separation", "target_evidence": "green recycling icon", "expected_visual_elements": ["dashboard panel", "green recycling icon"], "negative_prompt": "low detail, blurry, background clutter, text labels", "diversity_tags": ["dashboard panel", "recycling icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "d1f3728e53c1", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard panel with", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_d1f3728e53c1", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_d1f3728e53c1/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29159854, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_d1f3728e53c1/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_d1f3728e53c1/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_d1f3728e53c1/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4980542, "qwen_elapsed_seconds": 5.241, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_d1f3728e53c1/record.json"} +{"prompt_id": "visual_knowledge_d3_e7e9713c4122", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic box with a fork and knife icon on a light gray surface, high contrast, clean boundaries", "target_evidence": "fork and knife icon", "expected_visual_elements": ["box", "fork and knife icon"], "negative_prompt": "distorted, low contrast, background distractions, watermarks", "diversity_tags": ["box", "food icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e7e9713c4122", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic box with a", "max_similarity_to_other_prompt": 0.7143, "diversity_score": 0.0857, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_e7e9713c4122", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_e7e9713c4122/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22170593, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_e7e9713c4122/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_e7e9713c4122/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_e7e9713c4122/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1423661, "qwen_elapsed_seconds": 5.246, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_e7e9713c4122/record.json"} +{"prompt_id": "visual_knowledge_d3_eb55aa20a228", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic control panel with a yellow lightning bolt icon on a dark background, clean and sharp, isolated for symbol analysis", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["control panel", "yellow lightning bolt icon"], "negative_prompt": "blurry, low contrast, overlapping elements, complex background", "diversity_tags": ["control panel", "lightning hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "eb55aa20a228", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic control panel with", "max_similarity_to_other_prompt": 0.6154, "diversity_score": 0.1846, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_eb55aa20a228", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_eb55aa20a228/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 29429995, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_eb55aa20a228/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_eb55aa20a228/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_eb55aa20a228/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2044464, "qwen_elapsed_seconds": 5.182, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_eb55aa20a228/record.json"} +{"prompt_id": "visual_knowledge_d3_f4d3acddb194", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 3, "image_prompt": "synthetic photorealistic equipment tag with a skull warning icon, placed on a metal surface, clear foreground, high contrast", "target_evidence": "skull warning icon", "expected_visual_elements": ["equipment tag", "skull warning icon"], "negative_prompt": "faded, distorted, text labels, background clutter", "diversity_tags": ["equipment tag", "hazard icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "f4d3acddb194", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment tag with", "max_similarity_to_other_prompt": 0.75, "diversity_score": 0.05, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d3_f4d3acddb194", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d3_f4d3acddb194/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23623397, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d3_f4d3acddb194/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_f4d3acddb194/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d3_f4d3acddb194/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2229475, "qwen_elapsed_seconds": 5.249, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d3_f4d3acddb194/record.json"} +{"prompt_id": "visual_knowledge_d4_04faedb1ba04", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic box with a wheelchair accessibility icon on a light gray box, clean background, high contrast, clear boundaries", "target_evidence": "wheelchair accessibility icon", "expected_visual_elements": ["box", "icon"], "negative_prompt": "distorted, low detail, overlapping, complex background", "diversity_tags": ["box", "accessibility icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "04faedb1ba04", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic box with a", "max_similarity_to_other_prompt": 0.619, "diversity_score": 0.181, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_04faedb1ba04", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_04faedb1ba04/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23826092, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_04faedb1ba04/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_04faedb1ba04/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_04faedb1ba04/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3427863, "qwen_elapsed_seconds": 5.156, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_04faedb1ba04/record.json"} +{"prompt_id": "visual_knowledge_d4_0f7bc5f50caa", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic notice board in a hospital corridor with a red medical cross icon clearly visible against a white background, high contrast and sharp focus for easy decomposition", "target_evidence": "red medical cross", "expected_visual_elements": ["notice board", "red medical cross icon", "white background", "sharp focus"], "negative_prompt": "low contrast, blurry edges, overlapping objects, poor lighting, watermarks, real brand logos", "diversity_tags": ["notice board", "medical icon", "hospital setting"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0f7bc5f50caa", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notice board in", "max_similarity_to_other_prompt": 0.5484, "diversity_score": 0.4516, "repeated_opening": false, "review": {"prompt_id": "visual_knowledge_d4_0f7bc5f50caa", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_0f7bc5f50caa/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24276163, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_0f7bc5f50caa/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_0f7bc5f50caa/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_0f7bc5f50caa/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2289023, "qwen_elapsed_seconds": 5.213, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_0f7bc5f50caa/record.json"} +{"prompt_id": "visual_knowledge_d4_1c3581c0c696", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic equipment label with a flame warning icon on a metallic surface, high contrast and clean edges", "target_evidence": "flame warning icon", "expected_visual_elements": ["equipment label", "flame warning icon"], "negative_prompt": "faded text, low detail, real branding", "diversity_tags": ["equipment label", "fire safety"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1c3581c0c696", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment label with", "max_similarity_to_other_prompt": 0.619, "diversity_score": 0.181, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_1c3581c0c696", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_1c3581c0c696/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21292967, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_1c3581c0c696/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_1c3581c0c696/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_1c3581c0c696/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9807308, "qwen_elapsed_seconds": 5.264, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_1c3581c0c696/record.json"} +{"prompt_id": "visual_knowledge_d4_28cd45baab0d", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic equipment label with a wifi icon on a metallic surface, high contrast, clear foreground, sharp edges", "target_evidence": "wifi icon", "expected_visual_elements": ["equipment label", "icon"], "negative_prompt": "faded, low detail, overlapping, complex textures", "diversity_tags": ["equipment label", "wifi icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "28cd45baab0d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment label with", "max_similarity_to_other_prompt": 0.619, "diversity_score": 0.181, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_28cd45baab0d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_28cd45baab0d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22818956, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_28cd45baab0d/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_28cd45baab0d/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_28cd45baab0d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2193169, "qwen_elapsed_seconds": 5.195, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_28cd45baab0d/record.json"} +{"prompt_id": "visual_knowledge_d4_38d09579b5b3", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic wall sticker showing a red medical cross with clean edges and bright colors against a plain white wall", "target_evidence": "red medical cross", "expected_visual_elements": ["wall sticker", "red medical cross"], "negative_prompt": "distracting patterns, low resolution, real logos", "diversity_tags": ["wall sticker", "medical aid"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "38d09579b5b3", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wall sticker showing", "max_similarity_to_other_prompt": 0.4583, "diversity_score": 0.3417, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_38d09579b5b3", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_38d09579b5b3/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28337759, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_38d09579b5b3/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_38d09579b5b3/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_38d09579b5b3/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2143177, "qwen_elapsed_seconds": 5.163, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_38d09579b5b3/record.json"} +{"prompt_id": "visual_knowledge_d4_4b4a7d38b8e0", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic control panel with a yellow lightning bolt icon on a dark background, sharp focus on the symbol, clear boundaries, and high contrast", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["control panel", "yellow lightning bolt icon"], "negative_prompt": "blurry, low contrast, real brand logos, watermarks", "diversity_tags": ["control panel", "electrical hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "4b4a7d38b8e0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic control panel with", "max_similarity_to_other_prompt": 0.68, "diversity_score": 0.12, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_4b4a7d38b8e0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_4b4a7d38b8e0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27304158, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_4b4a7d38b8e0/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_4b4a7d38b8e0/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_4b4a7d38b8e0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3978753, "qwen_elapsed_seconds": 5.259, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_4b4a7d38b8e0/record.json"} +{"prompt_id": "visual_knowledge_d4_5430a0e33d62", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic dashboard panel with a yellow lightning bolt warning sign on a clean white background, sharp focus, clear foreground separation", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["dashboard panel", "warning sign"], "negative_prompt": "blurry, low contrast, overlapping elements, complex background", "diversity_tags": ["dashboard", "warning sign"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5430a0e33d62", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard panel with", "max_similarity_to_other_prompt": 0.7273, "diversity_score": 0.0727, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_5430a0e33d62", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_5430a0e33d62/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24593338, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_5430a0e33d62/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_5430a0e33d62/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_5430a0e33d62/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9197487, "qwen_elapsed_seconds": 5.125, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_5430a0e33d62/record.json"} +{"prompt_id": "visual_knowledge_d4_681df27f690c", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic notice board with a leaf eco icon on a wooden panel, bright colors, and high contrast for easy identification", "target_evidence": "leaf eco icon", "expected_visual_elements": ["notice board", "leaf eco icon"], "negative_prompt": "faded text, real branding, low resolution", "diversity_tags": ["notice board", "eco-friendly"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "681df27f690c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notice board with", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_681df27f690c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_681df27f690c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28187964, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_681df27f690c/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_681df27f690c/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_681df27f690c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8646018, "qwen_elapsed_seconds": 5.24, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_681df27f690c/record.json"} +{"prompt_id": "visual_knowledge_d4_70a6f85c77f7", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic equipment tag with a skull warning icon partially covered by a label, high contrast and sharp focus on the icon", "target_evidence": "skull warning icon", "expected_visual_elements": ["equipment tag", "skull warning icon"], "negative_prompt": "faded colors, low detail, real branding", "diversity_tags": ["equipment tag", "toxic hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "70a6f85c77f7", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment tag with", "max_similarity_to_other_prompt": 0.5385, "diversity_score": 0.2615, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_70a6f85c77f7", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_70a6f85c77f7/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27775893, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_70a6f85c77f7/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_70a6f85c77f7/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_70a6f85c77f7/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3621725, "qwen_elapsed_seconds": 5.162, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_70a6f85c77f7/record.json"} +{"prompt_id": "visual_knowledge_d4_795bc8b4e673", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic equipment tag showing a skull warning icon on a dark gray surface, sharp focus, clear separation from surrounding text", "target_evidence": "skull warning icon", "expected_visual_elements": ["equipment tag", "icon"], "negative_prompt": "low detail, overlapping text, blurred icon, complex background", "diversity_tags": ["equipment tag", "hazard icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "795bc8b4e673", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment tag showing", "max_similarity_to_other_prompt": 0.4231, "diversity_score": 0.3769, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_795bc8b4e673", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_795bc8b4e673/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22344015, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_795bc8b4e673/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_795bc8b4e673/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_795bc8b4e673/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5143072, "qwen_elapsed_seconds": 5.228, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_795bc8b4e673/record.json"} +{"prompt_id": "visual_knowledge_d4_7c65aae459cb", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic control panel with a green recycling icon on a metallic surface, clear foreground, sharp edges, high definition", "target_evidence": "green recycling icon", "expected_visual_elements": ["control panel", "icon"], "negative_prompt": "low resolution, occluded icon, complex textures, distorted", "diversity_tags": ["control panel", "eco icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "7c65aae459cb", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic control panel with", "max_similarity_to_other_prompt": 0.5909, "diversity_score": 0.2091, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_7c65aae459cb", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_7c65aae459cb/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21929967, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_7c65aae459cb/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_7c65aae459cb/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_7c65aae459cb/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7935839, "qwen_elapsed_seconds": 5.248, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_7c65aae459cb/record.json"} +{"prompt_id": "visual_knowledge_d4_8789b624fbc4", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic panel with a fork and knife icon on a light beige surface, clean boundaries, high contrast, clear foreground", "target_evidence": "fork and knife icon", "expected_visual_elements": ["panel", "icon"], "negative_prompt": "low resolution, distorted, overlapping, complex background", "diversity_tags": ["panel", "food icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "8789b624fbc4", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic panel with a", "max_similarity_to_other_prompt": 0.7143, "diversity_score": 0.0857, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_8789b624fbc4", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_8789b624fbc4/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24676021, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_8789b624fbc4/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_8789b624fbc4/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_8789b624fbc4/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7295496, "qwen_elapsed_seconds": 5.295, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_8789b624fbc4/record.json"} +{"prompt_id": "visual_knowledge_d4_99c9030ca1d7", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic wall sticker with a water droplet icon on a light blue wall, clean boundaries, high contrast between icon and background", "target_evidence": "water droplet icon", "expected_visual_elements": ["wall sticker", "icon"], "negative_prompt": "faded, low contrast, overlapping elements, complex textures", "diversity_tags": ["wall sticker", "water icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "99c9030ca1d7", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wall sticker with", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_99c9030ca1d7", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_99c9030ca1d7/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22100006, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_99c9030ca1d7/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_99c9030ca1d7/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_99c9030ca1d7/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6012520, "qwen_elapsed_seconds": 5.187, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_99c9030ca1d7/record.json"} +{"prompt_id": "visual_knowledge_d4_b1c642b1f448", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic notice board displaying a red medical cross icon with a simple white border, high contrast, clean boundaries", "target_evidence": "red medical cross", "expected_visual_elements": ["notice board", "icon"], "negative_prompt": "distorted, low resolution, occluded icon, complex patterns", "diversity_tags": ["notice board", "medical icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b1c642b1f448", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notice board displaying", "max_similarity_to_other_prompt": 0.5833, "diversity_score": 0.2167, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_b1c642b1f448", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_b1c642b1f448/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28850346, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_b1c642b1f448/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_b1c642b1f448/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_b1c642b1f448/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3654558, "qwen_elapsed_seconds": 5.224, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_b1c642b1f448/record.json"} +{"prompt_id": "visual_knowledge_d4_b2d24549fb8a", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic dashboard panel showing a water droplet icon with sharp focus and clean lines on a dark interface", "target_evidence": "water droplet icon", "expected_visual_elements": ["dashboard panel", "water droplet icon"], "negative_prompt": "overlapping elements, real logos, low resolution", "diversity_tags": ["dashboard panel", "water source"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b2d24549fb8a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard panel showing", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_b2d24549fb8a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_b2d24549fb8a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 22741229, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_b2d24549fb8a/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_b2d24549fb8a/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_b2d24549fb8a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7129710, "qwen_elapsed_seconds": 5.21, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_b2d24549fb8a/record.json"} +{"prompt_id": "visual_knowledge_d4_da08ebda1180", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic panel with a wifi icon on a light-colored surface, sharp focus and clean boundaries for easy decomposition", "target_evidence": "wifi icon", "expected_visual_elements": ["panel", "wifi icon"], "negative_prompt": "overlapping elements, low resolution, real logos", "diversity_tags": ["panel", "wireless access"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "da08ebda1180", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic panel with a", "max_similarity_to_other_prompt": 0.5909, "diversity_score": 0.2091, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_da08ebda1180", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_da08ebda1180/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 24493577, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_da08ebda1180/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_da08ebda1180/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_da08ebda1180/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2507275, "qwen_elapsed_seconds": 5.243, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_da08ebda1180/record.json"} +{"prompt_id": "visual_knowledge_d4_e967c6faa2f5", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic box with a fork and knife icon on its side, clear boundaries and high contrast for visual separation", "target_evidence": "fork and knife icon", "expected_visual_elements": ["box", "fork and knife icon"], "negative_prompt": "low contrast, real logos, blurred edges", "diversity_tags": ["box", "food service"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e967c6faa2f5", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic box with a", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_e967c6faa2f5", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_e967c6faa2f5/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20966840, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_e967c6faa2f5/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_e967c6faa2f5/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_e967c6faa2f5/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 396008, "qwen_elapsed_seconds": 5.175, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_e967c6faa2f5/record.json"} +{"prompt_id": "visual_knowledge_d4_ee288d46259e", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic dashboard showing a leaf eco icon with a clean white border, sharp focus, clear separation from other controls", "target_evidence": "leaf eco icon", "expected_visual_elements": ["dashboard", "icon"], "negative_prompt": "low contrast, blurred, overlapping elements, complex background", "diversity_tags": ["dashboard", "eco icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ee288d46259e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard showing a", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_ee288d46259e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_ee288d46259e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25452887, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_ee288d46259e/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_ee288d46259e/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_ee288d46259e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1103685, "qwen_elapsed_seconds": 5.223, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_ee288d46259e/record.json"} +{"prompt_id": "visual_knowledge_d4_ff6c8428efa6", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic dashboard with a green recycling icon on a dark panel, clean lines, and clear separation from other controls", "target_evidence": "green recycling icon", "expected_visual_elements": ["dashboard", "green recycling icon"], "negative_prompt": "overlapping elements, low resolution, real logos", "diversity_tags": ["dashboard", "recycling"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ff6c8428efa6", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard with a", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_ff6c8428efa6", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_ff6c8428efa6/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25863434, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_ff6c8428efa6/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_ff6c8428efa6/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_ff6c8428efa6/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4130803, "qwen_elapsed_seconds": 5.221, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_ff6c8428efa6/record.json"} +{"prompt_id": "visual_knowledge_d4_ff91053284ce", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 4, "image_prompt": "synthetic photorealistic folder with a flame warning icon on a light yellow background, sharp focus, clear separation, high definition", "target_evidence": "flame warning icon", "expected_visual_elements": ["folder", "icon"], "negative_prompt": "blurry, low contrast, occluded, complex background", "diversity_tags": ["folder", "fire icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "ff91053284ce", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic folder with a", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d4_ff91053284ce", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d4_ff91053284ce/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23448951, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d4_ff91053284ce/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_ff91053284ce/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d4_ff91053284ce/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 5257932, "qwen_elapsed_seconds": 5.148, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d4_ff91053284ce/record.json"} +{"prompt_id": "visual_knowledge_d5_03f61bda150c", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic dashboard with a water droplet icon on a gray panel, clean lines, high contrast, isolated symbol in a cluttered scene", "target_evidence": "water droplet icon", "expected_visual_elements": ["dashboard", "gray panel", "blue symbol"], "negative_prompt": "low resolution, unclear boundaries, overlapping text", "diversity_tags": ["dashboard", "water source"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "03f61bda150c", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard with a", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_03f61bda150c", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_03f61bda150c/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23797236, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_03f61bda150c/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_03f61bda150c/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_03f61bda150c/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3302387, "qwen_elapsed_seconds": 5.203, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_03f61bda150c/record.json"} +{"prompt_id": "visual_knowledge_d5_0cea028a8843", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic control panel with a fork and knife icon on a clean metal surface, next to a food tray and utensils", "target_evidence": "fork and knife icon", "expected_visual_elements": ["control panel", "food service icon", "metal surface", "food tray", "utensils"], "negative_prompt": "low resolution, overlapping elements, unrealistic colors", "diversity_tags": ["control panel", "food service icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0cea028a8843", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic control panel with", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_0cea028a8843", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_0cea028a8843/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26130375, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_0cea028a8843/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_0cea028a8843/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_0cea028a8843/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4035333, "qwen_elapsed_seconds": 5.238, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_0cea028a8843/record.json"} +{"prompt_id": "visual_knowledge_d5_0d0c8fe76555", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic wall sticker with a wheelchair accessibility icon on a light blue wall, sharply focused and separated from surrounding elements", "target_evidence": "wheelchair accessibility icon", "expected_visual_elements": ["wall sticker", "light blue wall", "icon"], "negative_prompt": "blurry, low contrast, overlapping text, cartoon style", "diversity_tags": ["wall sticker", "accessibility"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "0d0c8fe76555", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wall sticker with", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_0d0c8fe76555", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_0d0c8fe76555/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20933228, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_0d0c8fe76555/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_0d0c8fe76555/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_0d0c8fe76555/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3214732, "qwen_elapsed_seconds": 5.143, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_0d0c8fe76555/record.json"} +{"prompt_id": "visual_knowledge_d5_1702d3424ba0", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic dashboard with a yellow lightning bolt warning icon on a clean panel, surrounded by digital meters and buttons", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["dashboard", "warning icon", "digital meters", "buttons"], "negative_prompt": "blurry, low contrast, unrealistic lighting, text on the icon", "diversity_tags": ["dashboard", "warning icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "1702d3424ba0", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard with a", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_1702d3424ba0", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_1702d3424ba0/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23649173, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_1702d3424ba0/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_1702d3424ba0/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_1702d3424ba0/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4775723, "qwen_elapsed_seconds": 5.182, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_1702d3424ba0/record.json"} +{"prompt_id": "visual_knowledge_d5_38929646b03e", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic folder with a wifi icon on a light yellow background, sharply focused and visually separated from other items", "target_evidence": "wifi icon", "expected_visual_elements": ["folder", "light yellow background", "icon"], "negative_prompt": "low resolution, cluttered background, cartoon style, watermarks", "diversity_tags": ["folder", "wireless access"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "38929646b03e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic folder with a", "max_similarity_to_other_prompt": 0.5652, "diversity_score": 0.2348, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_38929646b03e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_38929646b03e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26690434, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_38929646b03e/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_38929646b03e/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_38929646b03e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6944996, "qwen_elapsed_seconds": 5.191, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_38929646b03e/record.json"} +{"prompt_id": "visual_knowledge_d5_48dc1f63e752", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic panel with a fork and knife icon on a light gray surface, high contrast and clear boundaries for easy decomposition", "target_evidence": "fork and knife icon", "expected_visual_elements": ["panel", "light gray surface", "icon"], "negative_prompt": "blurry, low contrast, overlapping text, cartoon style", "diversity_tags": ["panel", "food service"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "48dc1f63e752", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic panel with a", "max_similarity_to_other_prompt": 0.6957, "diversity_score": 0.1043, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_48dc1f63e752", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_48dc1f63e752/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27552419, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_48dc1f63e752/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_48dc1f63e752/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_48dc1f63e752/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8875726, "qwen_elapsed_seconds": 5.251, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_48dc1f63e752/record.json"} +{"prompt_id": "visual_knowledge_d5_5841699e603e", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic notice board with a water droplet icon in a clear frame, displayed in a public rest area with a bench and trash can", "target_evidence": "water droplet icon", "expected_visual_elements": ["notice board", "water icon", "public rest area", "bench", "trash can"], "negative_prompt": "low contrast, blurred icon, text on the icon", "diversity_tags": ["notice board", "water icon"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "5841699e603e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notice board with", "max_similarity_to_other_prompt": 0.3448, "diversity_score": 0.4552, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_5841699e603e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_5841699e603e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25645735, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_5841699e603e/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_5841699e603e/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_5841699e603e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 3071940, "qwen_elapsed_seconds": 5.28, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_5841699e603e/record.json"} +{"prompt_id": "visual_knowledge_d5_6241c34e055a", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic box with a leaf eco icon on a green surface, high contrast and clearly separated from surrounding objects", "target_evidence": "leaf eco icon", "expected_visual_elements": ["box", "green surface", "icon"], "negative_prompt": "blurry, low contrast, overlapping elements, cartoon style", "diversity_tags": ["box", "eco-friendly"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "6241c34e055a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic box with a", "max_similarity_to_other_prompt": 0.5, "diversity_score": 0.3, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_6241c34e055a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_6241c34e055a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21531911, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_6241c34e055a/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_6241c34e055a/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_6241c34e055a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9323587, "qwen_elapsed_seconds": 6.269, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_6241c34e055a/record.json"} +{"prompt_id": "visual_knowledge_d5_696807b2276a", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic wall sticker with a wheelchair accessibility icon on a light green wall, sharp focus, clear foreground separation", "target_evidence": "wheelchair accessibility icon", "expected_visual_elements": ["wall sticker", "light green wall", "blue symbol"], "negative_prompt": "low detail, poor lighting, unclear boundaries", "diversity_tags": ["wall sticker", "accessibility"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "696807b2276a", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic wall sticker with", "max_similarity_to_other_prompt": 0.619, "diversity_score": 0.181, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_696807b2276a", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_696807b2276a/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 28774332, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_696807b2276a/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_696807b2276a/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_696807b2276a/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8015722, "qwen_elapsed_seconds": 5.217, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_696807b2276a/record.json"} +{"prompt_id": "visual_knowledge_d5_801486d83d5d", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic equipment label with a flame warning icon on a red background, sharp focus and clean edges for easy decomposition", "target_evidence": "flame warning icon", "expected_visual_elements": ["equipment label", "red background", "flame icon"], "negative_prompt": "low detail, overlapping text, cartoon style, watermarks", "diversity_tags": ["equipment label", "fire safety"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "801486d83d5d", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment label with", "max_similarity_to_other_prompt": 0.6667, "diversity_score": 0.1333, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_801486d83d5d", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_801486d83d5d/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26032560, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_801486d83d5d/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_801486d83d5d/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_801486d83d5d/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4355955, "qwen_elapsed_seconds": 6.827, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_801486d83d5d/record.json"} +{"prompt_id": "visual_knowledge_d5_80ba8f9f78fc", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic dashboard with a water droplet icon on a dark blue panel, clearly visible and separated from other controls", "target_evidence": "water droplet icon", "expected_visual_elements": ["dashboard", "dark blue panel", "icon"], "negative_prompt": "blurry, low contrast, overlapping elements, cartoon style", "diversity_tags": ["dashboard", "water source"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "80ba8f9f78fc", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard with a", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_80ba8f9f78fc", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_80ba8f9f78fc/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 20416310, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_80ba8f9f78fc/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_80ba8f9f78fc/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_80ba8f9f78fc/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 7594692, "qwen_elapsed_seconds": 6.817, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_80ba8f9f78fc/record.json"} +{"prompt_id": "visual_knowledge_d5_8217f6b26930", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic notice board with a red medical cross symbol on a white background, clear foreground separation, high contrast, clean boundaries", "target_evidence": "red medical cross", "expected_visual_elements": ["notice board", "white background", "red symbol"], "negative_prompt": "blurry, low contrast, overlapping elements, poor lighting", "diversity_tags": ["notice board", "medical aid"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "8217f6b26930", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notice board with", "max_similarity_to_other_prompt": 0.5833, "diversity_score": 0.2167, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_8217f6b26930", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_8217f6b26930/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26125461, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_8217f6b26930/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_8217f6b26930/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_8217f6b26930/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 6476861, "qwen_elapsed_seconds": 6.821, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_8217f6b26930/record.json"} +{"prompt_id": "visual_knowledge_d5_8bf6ca2d1995", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic notice board displaying a skull warning icon on a light green background, high contrast and clean boundaries", "target_evidence": "skull warning icon", "expected_visual_elements": ["notice board", "light green background", "skull icon"], "negative_prompt": "low resolution, cluttered background, cartoon style, watermarks", "diversity_tags": ["notice board", "toxic hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "8bf6ca2d1995", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic notice board displaying", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_8bf6ca2d1995", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_8bf6ca2d1995/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 25444114, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_8bf6ca2d1995/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_8bf6ca2d1995/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_8bf6ca2d1995/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4799712, "qwen_elapsed_seconds": 5.768, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_8bf6ca2d1995/record.json"} +{"prompt_id": "visual_knowledge_d5_a05cf32f35ef", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic equipment tag with a yellow lightning bolt symbol on a yellow background, high contrast for clear visibility", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["equipment tag", "yellow background", "lightning bolt"], "negative_prompt": "low detail, overlapping text, cartoon style, watermarks", "diversity_tags": ["equipment tag", "electrical hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "a05cf32f35ef", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment tag with", "max_similarity_to_other_prompt": 0.5217, "diversity_score": 0.2783, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_a05cf32f35ef", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_a05cf32f35ef/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27032504, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_a05cf32f35ef/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_a05cf32f35ef/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_a05cf32f35ef/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2365821, "qwen_elapsed_seconds": 6.213, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_a05cf32f35ef/record.json"} +{"prompt_id": "visual_knowledge_d5_adb2f532e843", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic dashboard panel showing a green recycling icon against a dark gray surface, clearly visible with clean boundaries", "target_evidence": "green recycling icon", "expected_visual_elements": ["dashboard panel", "dark gray surface", "green icon"], "negative_prompt": "low resolution, cluttered background, cartoon style, watermarks", "diversity_tags": ["dashboard", "eco-friendly"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "adb2f532e843", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard panel showing", "max_similarity_to_other_prompt": 0.4615, "diversity_score": 0.3385, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_adb2f532e843", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_adb2f532e843/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21951075, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_adb2f532e843/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_adb2f532e843/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_adb2f532e843/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4632747, "qwen_elapsed_seconds": 6.235, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_adb2f532e843/record.json"} +{"prompt_id": "visual_knowledge_d5_b3dca94f7f5e", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic control panel with a red medical cross symbol on a clean white background, high contrast and sharp focus for easy decomposition", "target_evidence": "red medical cross", "expected_visual_elements": ["control panel", "white background", "red cross symbol"], "negative_prompt": "blurry, low contrast, overlapping elements, cartoon style, watermarks", "diversity_tags": ["control panel", "medical aid"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "b3dca94f7f5e", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic control panel with", "max_similarity_to_other_prompt": 0.72, "diversity_score": 0.08, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_b3dca94f7f5e", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_b3dca94f7f5e/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 26969791, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_b3dca94f7f5e/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_b3dca94f7f5e/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_b3dca94f7f5e/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 9035037, "qwen_elapsed_seconds": 6.327, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_b3dca94f7f5e/record.json"} +{"prompt_id": "visual_knowledge_d5_c1d6a336de09", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic control panel with a skull warning icon on a red panel, high contrast, clean separation, isolated symbol", "target_evidence": "skull warning icon", "expected_visual_elements": ["control panel", "red background", "black symbol"], "negative_prompt": "blurry, low contrast, overlapping text", "diversity_tags": ["control panel", "toxic hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "c1d6a336de09", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic control panel with", "max_similarity_to_other_prompt": 0.48, "diversity_score": 0.32, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_c1d6a336de09", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_c1d6a336de09/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 21717755, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_c1d6a336de09/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_c1d6a336de09/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_c1d6a336de09/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 4017449, "qwen_elapsed_seconds": 6.175, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_c1d6a336de09/record.json"} +{"prompt_id": "visual_knowledge_d5_e55db88fe1c9", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic dashboard panel showing a yellow lightning bolt icon against a dark panel, sharp focus, clean edges, isolated symbol", "target_evidence": "yellow lightning bolt", "expected_visual_elements": ["dashboard panel", "dark background", "yellow icon"], "negative_prompt": "distracting background, unclear boundaries, low resolution", "diversity_tags": ["dashboard panel", "electrical hazard"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "e55db88fe1c9", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic dashboard panel showing", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_e55db88fe1c9", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_e55db88fe1c9/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 27456505, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_e55db88fe1c9/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_e55db88fe1c9/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_e55db88fe1c9/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 1764192, "qwen_elapsed_seconds": 6.203, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_e55db88fe1c9/record.json"} +{"prompt_id": "visual_knowledge_d5_fa5219d618a2", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic equipment tag with a green recycling icon on a blue tag, clean lines, high contrast, clear separation from surroundings", "target_evidence": "green recycling icon", "expected_visual_elements": ["equipment tag", "blue background", "green symbol"], "negative_prompt": "overlapping text, low detail, poor lighting", "diversity_tags": ["equipment tag", "recycling"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "fa5219d618a2", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic equipment tag with", "max_similarity_to_other_prompt": 0.52, "diversity_score": 0.28, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_fa5219d618a2", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_fa5219d618a2/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 23944696, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_fa5219d618a2/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_fa5219d618a2/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_fa5219d618a2/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 2132575, "qwen_elapsed_seconds": 6.335, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_fa5219d618a2/record.json"} +{"prompt_id": "visual_knowledge_d5_fdd48c9a2b78", "run_name": "flux2_qil_decomposition_pilot", "category": "visual_knowledge", "difficulty": 5, "image_prompt": "synthetic photorealistic folder with a leaf eco icon on the front, sharp focus, clean edges, isolated symbol in a cluttered but readable scene", "target_evidence": "leaf eco icon", "expected_visual_elements": ["folder", "front panel", "green symbol"], "negative_prompt": "low resolution, unclear boundaries, overlapping objects", "diversity_tags": ["folder", "eco-friendly"], "llm_model_id": "Qwen/Qwen3-8B", "prompt_hash": "fdd48c9a2b78", "rule_valid": true, "rule_issues": [], "rule_score": 1.0, "opening_key": "synthetic photorealistic folder with a", "max_similarity_to_other_prompt": 0.4, "diversity_score": 0.4, "repeated_opening": true, "review": {"prompt_id": "visual_knowledge_d5_fdd48c9a2b78", "accept": true, "quality": 5, "decomposition_fit": 5, "diversity": 4, "issues": []}, "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", "final_rank_score": 0.97, "status": "qwen_layered_done", "image_model_id": "black-forest-labs/FLUX.2-dev", "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", "image_model_slug": "black-forest-labs_FLUX.2-dev", "image_path": "images/black-forest-labs_FLUX.2-dev/visual_knowledge/visual_knowledge_d5_fdd48c9a2b78/image.png", "image_width": 1024, "image_height": 1024, "num_inference_steps": 28, "guidance_scale": 3.5, "seed": 30086616, "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", "qwen_layer_count_requested": 3, "qwen_layer_count_returned": 3, "qwen_layer_paths": ["layers-x/k3/visual_knowledge/visual_knowledge_d5_fdd48c9a2b78/layers/layer_00.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_fdd48c9a2b78/layers/layer_01.png", "layers-x/k3/visual_knowledge/visual_knowledge_d5_fdd48c9a2b78/layers/layer_02.png"], "qwen_resolution": 640, "qwen_num_inference_steps": 5, "qwen_true_cfg_scale": 4.0, "qwen_seed": 8354082, "qwen_elapsed_seconds": 6.389, "qwen_record_path": "layers-x/k3/visual_knowledge/visual_knowledge_d5_fdd48c9a2b78/record.json"} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..f54cfa27e79024bb8983ef040eaaf95d2f98831d --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:003e31a118d5cbc1d1afe3e8bd832797ef91698c639aaeb7bf179e9ca572d66e +size 53258 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..133ab49bc83c6e2babb0a5b13267a3a2e17f24ae --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9d6a8e39de5ac068f4bdbef47ae8e021e976c99b424e8dc491c73251b23fdad5 +size 465906 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..c01b3e20cc23420a292e4debceb1fdc51d929141 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3269d24960dcc2431f36c0e5be44824f9311d7e78686faacf3ff1be36e9245a9 +size 469001 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/record.json new file mode 100644 index 0000000000000000000000000000000000000000..22d25fe8eead73e73bec73a510b6e7202972d3f7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_018f62776b1d", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "Synthetic photorealistic scene with a mannequin holding a key, clean boundaries and sharp focus on the key and hand, relation clearly visible", + "target_evidence": "mannequin holding key", + "expected_visual_elements": [ + "mannequin", + "key" + ], + "negative_prompt": "blurry, low contrast, overlapping objects", + "diversity_tags": [ + "holding", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "018f62776b1d", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.75, + "diversity_score": 0.05, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_018f62776b1d", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_018f62776b1d/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26286557, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5356689, + "qwen_elapsed_seconds": 5.173, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_018f62776b1d/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..07c59a982585b85813669c04169b10fb28912c34 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3525ddff2e884d46da86b3b83c0af2c5a81eb24a58717dbfabb1b3ab04eded6e +size 407969 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..9b8ea7ce1754770508f4ceec1121d8e811a66c83 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc137d0310fbca9ae6b7650767539ebb52b6abd6b3104033db58468fada36d88 +size 570029 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..97056ebe48f5c02a309e83466c3826393f2dd576 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:001a359804802611926eef62f1573d82dcd6d05fbf3242c2a3c54c4cb9e87794 +size 536298 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/record.json new file mode 100644 index 0000000000000000000000000000000000000000..17f4622cb5e4d957917e2d6acd28d741b9e84423 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_2056f61d59a6", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "Synthetic photorealistic view of a mannequin standing behind a desk, sharp focus and clear spatial relation between mannequin and desk", + "target_evidence": "mannequin behind desk", + "expected_visual_elements": [ + "mannequin", + "desk" + ], + "negative_prompt": "blurry, low depth, overlapping elements", + "diversity_tags": [ + "behind", + "desk" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "2056f61d59a6", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic view of a", + "max_similarity_to_other_prompt": 0.4545, + "diversity_score": 0.3455, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_2056f61d59a6", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_2056f61d59a6/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 21921962, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 8068497, + "qwen_elapsed_seconds": 5.156, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_2056f61d59a6/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..cf1289115ad455855d07cd7a3133a0061461dd7f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:437507d195c578338ffc5c046060648e667cbd3b2bb021c211cff088472c8f63 +size 218991 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..3c6ce5dcfee7ee3b8c65334adca5fa23b691b338 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d146247ccd1100a5b838f412d012962725d7c9ad3cd6ab4c41a7fe5b1e67fb95 +size 744254 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..7c90f26d9c65ae21b4a2e3c9754f3e9c394b6c23 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:604beaf9d588346215d18f62b3aefe6caf9c328a14ef2922620af55f92a46d8d +size 839348 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/record.json new file mode 100644 index 0000000000000000000000000000000000000000..f82b03b0b4211b49160f24aea799c9939459b956 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_26310eaeece0", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "synthetic photorealistic view of a bag attached to a backpack strap, clear foreground and sharp focus on the attachment point", + "target_evidence": "bag attached to backpack strap", + "expected_visual_elements": [ + "bag", + "backpack strap" + ], + "negative_prompt": "blurry, low contrast, watermarks, real people, faces", + "diversity_tags": [ + "attached to", + "backpack" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "26310eaeece0", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic view of a", + "max_similarity_to_other_prompt": 0.4167, + "diversity_score": 0.3833, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_26310eaeece0", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_26310eaeece0/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 27538671, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 4635157, + "qwen_elapsed_seconds": 5.209, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_26310eaeece0/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..6baeae1ac282876d334250e817420a253d5caacf --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:33afc310205ba7fa6ecf2f63dd7e1c71fdf3d5bbf19c928bc852b4c329e89cb5 +size 367346 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..8cdd483daa55f8bd1631c607de47b6a7e758f729 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f5fe4c4173a138bddfc160b0ae08d35905489743bc9d29370bedb0bfd84f13c +size 583572 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..77a349e00d181fdd86f6c4f4c442f8c28866048d --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02d41beb2e991b879dbc1f1fc2537ee29a2ab742c15f6990de8bd0beca3c9589 +size 613644 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/record.json new file mode 100644 index 0000000000000000000000000000000000000000..f611aff50c482312dbad6e58a1648f9cb8dd5b66 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/record.json @@ -0,0 +1,63 @@ +{ + "prompt_id": "multi_object_relations_d1_28810914b7b6", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "Synthetic photorealistic scene with a mannequin placing a bag on a table, clean boundaries and clear relation between mannequin and bag", + "target_evidence": "mannequin placing bag on table", + "expected_visual_elements": [ + "mannequin", + "bag", + "table" + ], + "negative_prompt": "low detail, out of focus, unclear positioning", + "diversity_tags": [ + "on top of", + "table" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "28810914b7b6", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.5238, + "diversity_score": 0.2762, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_28810914b7b6", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_28810914b7b6/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28738385, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 587097, + "qwen_elapsed_seconds": 5.172, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_28810914b7b6/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..9478622bc679d5b34c403a66f9608e71aa534fec --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9b322d362cb7257aad9d59e6ecb6c63628d775230a5cff4dc97a5622e08616c +size 574521 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..41e2ce52065858b011c0532ceb7a80a59a0bc0cb --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:68545b471901c84c3504b7b9b5a03dc128e08f3d64eb2d0858595ca58e54bdea +size 532913 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..8746b748430278d360eebce922b85aa8631a3a39 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:532c5e1f843ee603164dc80a736672bc570d503fdcd0125cab072399bb145e04 +size 770675 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/record.json new file mode 100644 index 0000000000000000000000000000000000000000..480120f2cf10527cc27e56f6c89b4466f1f90609 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_2c93806efe70", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "Synthetic photorealistic setup with a tool on top of a bag, high contrast and clear visual relation between tool and bag", + "target_evidence": "tool on top of bag", + "expected_visual_elements": [ + "tool", + "bag" + ], + "negative_prompt": "low resolution, unclear positioning, overlapping objects", + "diversity_tags": [ + "on top of", + "bag" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "2c93806efe70", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.6364, + "diversity_score": 0.1636, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_2c93806efe70", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_2c93806efe70/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 30003506, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7539450, + "qwen_elapsed_seconds": 5.155, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_2c93806efe70/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..c45bb7e23350c3965b3f40c01ca0b7dbef98f0ea --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a79f5a378c39eaf089c06a1ccafa008f3c18c108770d78ba0472636f5ae72b14 +size 436749 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..aa53bb86c29ad0fc95f93d9cff84e63722e9d38c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8a7e3ff5222a7dabd0635c07404c664209e859ac776712b109efdca9601af4b6 +size 591437 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..318c1ba3b0a67581a3bba110b283a4a08ad5fe4b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a38860dc7a63459027784c7e2546c93f457aeed5ccb820285561fd47cb1d9adc +size 577467 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/record.json new file mode 100644 index 0000000000000000000000000000000000000000..6c5501f01d79ac5eae2799b1c4402da21cdb168a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d1_431f8a0e9b3d", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "synthetic photorealistic scene with a mannequin holding a toolbox on a desk, clean boundaries and sharp focus for clear held-by relation", + "target_evidence": "mannequin holding toolbox", + "expected_visual_elements": [ + "mannequin", + "toolbox", + "desk" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, real people, faces", + "diversity_tags": [ + "holding", + "mannequin", + "desk" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "431f8a0e9b3d", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.6957, + "diversity_score": 0.1043, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_431f8a0e9b3d", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 30165319, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 405868, + "qwen_elapsed_seconds": 5.145, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_431f8a0e9b3d/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..b06579152ca6a38c83e46803233f5905807ccadd --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:974c157f93a282d1370905daf4d10c22cfbd5e77fe811ad8fc8e5fa33bb8ca6a +size 108944 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..e283d3597fab0583b865b3ac6af03fc93155bcce --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7cde8f526cf48d0efb81648a705dd4236cdc96da98c22c09cf5e499e723d78d4 +size 468202 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..fa58753f6d88bc14a73db9d9f6a72e2bef17e51c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57a6ddbc435d42379973217ab2a83920e223f87291815e2934c9ddcb68bba4d9 +size 543201 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/record.json new file mode 100644 index 0000000000000000000000000000000000000000..0221688d6c506d4a8b046f66f230c2dc7dbf8d59 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_4d1095976c42", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "Synthetic photorealistic setup with a box on top of a shelf, high contrast between box and shelf, relation visually legible", + "target_evidence": "box on shelf", + "expected_visual_elements": [ + "box", + "shelf" + ], + "negative_prompt": "low resolution, out of focus, cluttered background", + "diversity_tags": [ + "on top of", + "shelf" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "4d1095976c42", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.6667, + "diversity_score": 0.1333, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_4d1095976c42", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_4d1095976c42/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25329557, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1971817, + "qwen_elapsed_seconds": 5.207, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d1095976c42/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..0ac11daa2eb71b417141f21cc2aa4f34dabb9ff6 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e2b2902949d963aea44153b068e826c0b65de33b5ca238a1ee41ef129c73cd3a +size 401022 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..b9411fb03a1c40c0722e7c59947a1d9d3e6e66e8 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:54d81dc69cb044276e8bf2996e353e67aca17870cdf690f5476313d7b55c8093 +size 498977 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..760f23a5f0847ca3f6fe20d360f5bdb6fcc3bcec --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3a30c259e313c48603e84909f6ab744c6d78838c381e0beb618d3385a480b5b1 +size 627664 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/record.json new file mode 100644 index 0000000000000000000000000000000000000000..577d9b020247b603498673714236bf1f8fcc2f67 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_4d6b6f9003fb", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "Synthetic photorealistic image of a bag inside a container, clear foreground separation and sharp edges between bag and container", + "target_evidence": "bag inside container", + "expected_visual_elements": [ + "bag", + "container" + ], + "negative_prompt": "foggy, washed out colors, objects partially obscured", + "diversity_tags": [ + "inside", + "container" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "4d6b6f9003fb", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.5789, + "diversity_score": 0.2211, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_4d6b6f9003fb", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20849906, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5595332, + "qwen_elapsed_seconds": 5.273, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_4d6b6f9003fb/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..a5e844dabcfc7ef9ea372594bc1050986823f988 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff6125face7dcfcdc050c7255e362b8ee1a434ad18338ad41adb4e0fb35b88cb +size 462748 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..0bbfa1ff59464a8ed5fa54f46983c2bbbf91ac1f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c085f9d5f2af3baded81802e527ae11657a0d59e68f731d20423f5fcaaa4c55 +size 501170 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..dd03c81a3d671920b1ba4166045e458723df2fe1 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a64d568947bf022150b5488c4bbfab99ad35f38973c9b11c6936f56d506daaaf +size 539518 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/record.json new file mode 100644 index 0000000000000000000000000000000000000000..f6697bf34e1bfeab3344660fbe93e9a248616f13 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_528663b80801", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "Synthetic photorealistic scene with a tool attached to a sign, clean lines and clear relation between tool and sign", + "target_evidence": "tool attached to sign", + "expected_visual_elements": [ + "tool", + "sign" + ], + "negative_prompt": "distorted shapes, unclear connections, low detail", + "diversity_tags": [ + "attached-to", + "sign" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "528663b80801", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.65, + "diversity_score": 0.15, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_528663b80801", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_528663b80801/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 22459641, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1971405, + "qwen_elapsed_seconds": 5.085, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_528663b80801/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..5c600b50a6632e19aec0c7031cf5dd9b1406110b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a267b8a7a58a53316327b4cdf4193cc0e8cfd3193d5e412534a3bc677aa8c8f +size 468279 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..7b73e9cda774c7e774fc93aebff369b35e290be5 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5621b311af8e7baa32213313bad1d181c18a43a837fdd3b1e8aa512d3b2daa77 +size 556263 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..d754af04d0b0fc2e3184883e1286418c613c6f2a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a797247f927c5fcb44cdec738f3c546bb2225628e517b64d71c2d819ba829dd +size 555348 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/record.json new file mode 100644 index 0000000000000000000000000000000000000000..b7e9dc6a89c2b0920eb3fbb664402cc81e49bd8a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_52acc1c2fbb9", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "synthetic photorealistic setup with a tool under a container, clean separation and high contrast between the tool and container base", + "target_evidence": "tool under container", + "expected_visual_elements": [ + "tool", + "container" + ], + "negative_prompt": "blurry, low contrast, watermarks, real people, faces", + "diversity_tags": [ + "under", + "container" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "52acc1c2fbb9", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.6316, + "diversity_score": 0.1684, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_52acc1c2fbb9", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28233303, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9946804, + "qwen_elapsed_seconds": 5.154, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_52acc1c2fbb9/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..3593819436ff10e7714a9bfd580344b817886478 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:697f45fcc49bbe1db5cf5cc81d1831a87b29379403c44f387a747631348246b3 +size 380945 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..db52ff8cc637e4569f0716af6821a9cc287c62b9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f1b6d41bee986695eb573a03804934fe1be2db7b55153df88d9d6aaa89f84254 +size 623937 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..ce072bf969903d95cdf066ba8ba69c0ff1423916 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e84b3d872e6c6e8c4ed9d20caff01a0488a89426c4c7c78f6e14b7c4aa009a4 +size 567006 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/record.json new file mode 100644 index 0000000000000000000000000000000000000000..b6f7a6820aa0bd1a1d7418b3a1e13efdc74eefc8 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_586bda1348a2", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "Synthetic photorealistic scene with a mannequin holding a sign, clear foreground separation and sharp focus on the sign and hand", + "target_evidence": "mannequin holding sign", + "expected_visual_elements": [ + "mannequin", + "sign" + ], + "negative_prompt": "blurry, low contrast, overlapping objects", + "diversity_tags": [ + "holding", + "sign" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "586bda1348a2", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.8, + "diversity_score": 0.0, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_586bda1348a2", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_586bda1348a2/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20335950, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9694648, + "qwen_elapsed_seconds": 5.092, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_586bda1348a2/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..22e76829b72081868c02437d46df0cc527aaea78 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f8f33fa129def19b831da1b24f0be99e3a2ef53741b52635d8072f3163bec83 +size 373343 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..2402e371abcc195b6e8eef3b93982e80c209ad4b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9192258e9a56877b9279def91709ad2b06f71bd05315f707af4a9606cf7e2cc7 +size 581548 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..f5a72d08214c2a9a89290922c99de8073ec33b9d --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9da21cf6dbb3fcd2305a0819c80a973096a725814afe7f5acc70f2c5430afcef +size 613191 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/record.json new file mode 100644 index 0000000000000000000000000000000000000000..24aaccc71d00dc37e0118fd29814374ce98c33fe --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d1_6b46aeee2b20", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "synthetic photorealistic image of a mannequin with a tool attached to its arm, clear boundaries for attached-to relation", + "target_evidence": "tool attached to mannequin arm", + "expected_visual_elements": [ + "mannequin", + "tool", + "arm" + ], + "negative_prompt": "overlapping, low detail, real people, faces", + "diversity_tags": [ + "attached-to", + "mannequin", + "tool" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "6b46aeee2b20", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.5909, + "diversity_score": 0.2091, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_6b46aeee2b20", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29391804, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5613533, + "qwen_elapsed_seconds": 5.168, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_6b46aeee2b20/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..c19bfaf02f5d39018fcd789b56dd52100d4b670b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e3f77a41fb868d646834e62204c160ebab296e4b753205af654fda94556f104 +size 594158 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..1d93d6b57d7f1099eb19eb5e652217e3b7dd4226 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab88e412d53601ad60b3b5fe3b951f55a6d0d31db3b1461032e0e544b89a1687 +size 547095 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..e9017464e591147624bf99658194277a04651892 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f221cbdff809bf9395b067e77b4884235abf8d8afaed0290f081581904559372 +size 684808 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/record.json new file mode 100644 index 0000000000000000000000000000000000000000..0ef1975faa295da6d4b97c45681da41ba53fb783 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_80c1fe8f92a4", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "synthetic photorealistic image of a tool inside a toolbox, clean boundaries and high contrast between the tool and box interior", + "target_evidence": "tool inside toolbox", + "expected_visual_elements": [ + "tool", + "toolbox" + ], + "negative_prompt": "blurry, low contrast, watermarks, real people, faces", + "diversity_tags": [ + "inside", + "toolbox" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "80c1fe8f92a4", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.5238, + "diversity_score": 0.2762, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_80c1fe8f92a4", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20649079, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 4576292, + "qwen_elapsed_seconds": 5.139, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_80c1fe8f92a4/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..0398c85a919f2842b0c63531be9c649882fc5e9f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:087334b97622eaeab037d2aa5a3e7e69a9eedef607970dcd6be5ef5a116bc400 +size 386258 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..e43123cfc0ba5f3ef1c0b77803c309a95d7adedb --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:873326d9f2612f5f0edd84b7402d38d4ad93ba3b127cab5ddd7aa249d6da1100 +size 531928 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..5b8c09e10e71f4ae3eec823d5e2206965e891294 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:28ee4fe4cdc20655e02317e8e0463aae56f00e87bb0b003a02669bf22b243d7e +size 633265 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/record.json new file mode 100644 index 0000000000000000000000000000000000000000..ca580bf21471b56c95eb3d734dc75feef0de18aa --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d1_98504a155dad", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "synthetic photorealistic setup with a bag placed on top of a shelf, high contrast for visible on-top relation", + "target_evidence": "bag on top of shelf", + "expected_visual_elements": [ + "bag", + "shelf", + "shelf unit" + ], + "negative_prompt": "overlapping, low detail, real people, faces", + "diversity_tags": [ + "on-top", + "shelf", + "bag" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "98504a155dad", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.7222, + "diversity_score": 0.0778, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_98504a155dad", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_98504a155dad/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 21842447, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7329201, + "qwen_elapsed_seconds": 5.181, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_98504a155dad/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..fc189f1d72467aab4dfca625b0afaed2ef444a6a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:201f88f96da9cf66045e3af64b34f7b4b93a5c9a11f5eba0f9cf3b28452708b8 +size 425638 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..96409aee6fc9fb03efb1df5244e50b431f4bf01e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58f2590d436d6e91bb6d61d9f97569e39ec6b512afddae9cf929552cb3c81728 +size 485416 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..501db6351e13ea057ccf7892428752e2df95e400 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be3e8f41e32a38f5916a76350ef5f0f58d0318206fbd9e9b170eb45cffb0c553 +size 526736 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/record.json new file mode 100644 index 0000000000000000000000000000000000000000..fede974f2800f0c0e6f243d3ba58ca4f13270769 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_9ab06ddd4ac2", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "synthetic photorealistic setup with a box on top of a shelf, high contrast and clean separation between the box and shelf edges", + "target_evidence": "box on top of shelf", + "expected_visual_elements": [ + "box", + "shelf" + ], + "negative_prompt": "blurry, low contrast, watermarks, real people, faces", + "diversity_tags": [ + "on top of", + "shelf" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "9ab06ddd4ac2", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.8, + "diversity_score": 0.0, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_9ab06ddd4ac2", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28600316, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7036460, + "qwen_elapsed_seconds": 5.196, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_9ab06ddd4ac2/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..cbe1840297f2914b1088e57959c317b6d4a239b7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c3a19a3fb63383c5ad7194eba43e8f11d73ad153a42172d60d485d63cfdae500 +size 398527 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..c02f6671d27d51573cdf7274b789a28eab888b07 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6dc9fa53f0f208345b0875092fcd60307ca79fe4c0a375e47b713892aca6c18 +size 556613 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..29704ec73b5faa12687643a7dce6f2d83ec8d056 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2854f26c209565e74b9e752f583b4af428e583a1a5032df4daea4b3fa14edd38 +size 653846 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/record.json new file mode 100644 index 0000000000000000000000000000000000000000..0dcc81862bda05e07b59c114ee5b2f76fadb1e90 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_a5c5b8ab9db1", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "Synthetic photorealistic image of a key inside a box, sharp focus and clear visual separation between key and box", + "target_evidence": "key inside box", + "expected_visual_elements": [ + "key", + "box" + ], + "negative_prompt": "foggy, washed out colors, overlapping objects", + "diversity_tags": [ + "inside", + "box" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "a5c5b8ab9db1", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.5789, + "diversity_score": 0.2211, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_a5c5b8ab9db1", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24101894, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5309862, + "qwen_elapsed_seconds": 5.196, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_a5c5b8ab9db1/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..6919afbc502cc9a0113182473c11d60ba58a8261 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e7c408815de5d47ace3a54057472c8cba33f1a797e7f1391df0a819780976d84 +size 409689 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..b6f283d1c28d061a57fa41829fdb71b9df2c8fc5 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2d8f30d7c37f2ec8c12363075ba4989616708c15fd5579fd21b55a8ca91505e +size 512833 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..5a902ef4fba7ab57c3eb55a54c84dafd2f75122b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db3f0d16b72dd7734392ece2a47c3f327c8eff617a6014fa0f9d3f282c202ebf +size 449114 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/record.json new file mode 100644 index 0000000000000000000000000000000000000000..c4c4f3894f2b191ed20e96099c006fc39177e562 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_af1dae7eec23", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "synthetic photorealistic scene with a sign behind a desk, sharp focus and clear separation between the sign and desk surface", + "target_evidence": "sign behind desk", + "expected_visual_elements": [ + "sign", + "desk" + ], + "negative_prompt": "blurry, low contrast, watermarks, real people, faces", + "diversity_tags": [ + "behind", + "desk" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "af1dae7eec23", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.5714, + "diversity_score": 0.2286, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_af1dae7eec23", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_af1dae7eec23/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26233414, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 4368367, + "qwen_elapsed_seconds": 5.233, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_af1dae7eec23/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..37feb6d605919d3216794c7118943fea2ad836d2 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:784fb6b8706b3445727093e55acb0c8f40d2042bbba67d1339d95a5f51a24842 +size 393585 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..7006e35d48d618993cae837df65b5bffda3d4345 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0cf1d2e19b396034724214323eae20bb9e9a3f0315b08b0891c8f7cfa24d0c1a +size 585364 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..e9bd3edcd2e14a0eabc162b1cc19b82604ec94a1 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7324d2fd517e4861bd5ca06d2a4b959afb02763e77f9c0a1cfa381e211682ca1 +size 580222 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/record.json new file mode 100644 index 0000000000000000000000000000000000000000..744744d7eb424dd280dd8b2317a0a206e7399a7c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_b8bb45add71b", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "synthetic photorealistic scene with a mannequin holding a key, clean boundaries and sharp focus on the key and hand, clear foreground separation", + "target_evidence": "mannequin holding key", + "expected_visual_elements": [ + "mannequin", + "key" + ], + "negative_prompt": "blurry, low contrast, watermarks, real people, faces", + "diversity_tags": [ + "holding", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "b8bb45add71b", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.8, + "diversity_score": 0.0, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_b8bb45add71b", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_b8bb45add71b/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28969010, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6098582, + "qwen_elapsed_seconds": 5.228, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_b8bb45add71b/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..e6371297bae4e647bed9720f5b69d50b961393af --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4f9765482fb210008c91c20fc01197e4d4e40c43c69699af56288a1e0ef7447 +size 345370 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..e49a9dee158fd28b62a20a92104c948a58518819 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d6df0162b445798dcfec537b714949377c1d9b4b2f1554ce63e4df4a0a48a80 +size 622247 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..029cc0bb7bed6a58efdade78218f550c7c9db926 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a47045e71ee97696e5c568fb660db649887b78e639c9e24d1eed856d8e2c19a4 +size 530616 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/record.json new file mode 100644 index 0000000000000000000000000000000000000000..12c59e948459da416b6a1b3e2a491383c822f9ef --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d1_dd45a37dffac", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "Synthetic photorealistic setup with a tool under a container, high contrast and clear spatial relation between tool and container", + "target_evidence": "tool under container", + "expected_visual_elements": [ + "tool", + "container" + ], + "negative_prompt": "low contrast, ambiguous positioning, overlapping objects", + "diversity_tags": [ + "under", + "container" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "dd45a37dffac", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.6316, + "diversity_score": 0.1684, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_dd45a37dffac", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_dd45a37dffac/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20420612, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7009906, + "qwen_elapsed_seconds": 5.143, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_dd45a37dffac/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..ff3ada5857e99dc320fa67d7c3b0be995b9b3993 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c76622d75b9f568bfdbd44c6e996f56f52e97da2ca4bd77d467e3486de2a5ee7 +size 539581 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..23d44ea10c886cfedba7fa8d6566a228dbcc6521 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cccbcaed2d8f23231aeb4172bc32ca28a953b9ec7c64fd305612ada54c94afbd +size 468967 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..cc6d1bbce6dcf91b357c11f6a34c58a63ef98d4c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1daaddb000d73c32f6a007dd804d3f72681ec8a356282d59d375986edc87adb0 +size 492467 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/record.json new file mode 100644 index 0000000000000000000000000000000000000000..1d0809fff5c8588393f50d4735f50e35b3a5588a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d1_fb7d932b5707", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 1, + "image_prompt": "synthetic photorealistic scene with a container on top of a shelf, high contrast for visible on-top relation", + "target_evidence": "container on top of shelf", + "expected_visual_elements": [ + "container", + "shelf", + "shelf unit" + ], + "negative_prompt": "blurry, low detail, real people, faces", + "diversity_tags": [ + "on-top", + "shelf", + "container" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "fb7d932b5707", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.7222, + "diversity_score": 0.0778, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d1_fb7d932b5707", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d1_fb7d932b5707/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 21246195, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 4532813, + "qwen_elapsed_seconds": 5.164, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d1_fb7d932b5707/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..df0f6eb8e66dcf4119b06af357c94782905547a8 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3284ebcb1dbc7b211b230fea3f1d776c974b4e63edd74aa0bcf6aa1d63adab3c +size 524289 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..083bec9c34d5df32195995fdd956812933e3e3d9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6bdcd7f591639ade508bae9df1e72ce976b308c313a7095e3d9a0c9ab8fd727 +size 630761 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..af7ee571c078dc509948a641d876f78cb936edee --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ebfa77ac82702b177b027834cb36ec0402e3b50039d85d1d148bbbf6e1629d78 +size 554330 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/record.json new file mode 100644 index 0000000000000000000000000000000000000000..a8bedc401b8ccb0313de58c715c8ef0191580e08 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d2_0a361d43b600", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic scene with a mannequin holding a bag, clean boundaries and sharp focus on the hand and bag", + "target_evidence": "mannequin holding bag", + "expected_visual_elements": [ + "mannequin", + "bag" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "holding", + "bag" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "0a361d43b600", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.8333, + "diversity_score": 0.0, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_0a361d43b600", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_0a361d43b600/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 21501073, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 8016721, + "qwen_elapsed_seconds": 5.107, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_0a361d43b600/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..c5d2a6d42132a8a1e2c3dfb526a532064a93b0df --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:acbe7af7b51511e002914213f7266ddc0bf3eefc34e7b0ab702aae71a7f4d2ba +size 345616 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..0d51b0896d43902412d6aa02cc0656bc2c0ff153 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:063d7b471c9361b22d209e11a992c49ee5cf700b584c52a01f2743cbb481669a +size 892511 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..9d6e3dfa2c9c19e3b791c29031aebcfde2236e65 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a0d24fd749bcb1732aeb7d0045b23e6731d80566b40be4a7cbc77ab4771ea37f +size 556767 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/record.json new file mode 100644 index 0000000000000000000000000000000000000000..c3e8ae3fe9f7b9acfd51f3521896d652fa2b820d --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d2_0c527d58a238", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic scene with a sign attached to a wall, clear foreground separation and high contrast for visual inspection", + "target_evidence": "sign attached to wall", + "expected_visual_elements": [ + "sign", + "wall", + "clear foreground" + ], + "negative_prompt": "overlapping, low detail, unrealistic lighting, blurry edges", + "diversity_tags": [ + "attached to", + "sign", + "wall" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "0c527d58a238", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.75, + "diversity_score": 0.05, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_0c527d58a238", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_0c527d58a238/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 21172209, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9882009, + "qwen_elapsed_seconds": 5.223, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_0c527d58a238/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..2128da9e5b3bb609fbe3f2d5646da0211ae9b21b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b9e78b708b8fd98ab3a3fe37c60545703318d129374277518097523ae4e9351 +size 456292 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..c99f7763c0e3196c4ddd4765ccf2694caf258364 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41e608eba5c1ea2e481a7d28587e0b61ef38058a5697c0e551e38f0afe85205a +size 591385 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..691025f421fc088f261d4a2d3c4a16b51e8c4ee4 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:28d285d82d03c00e795d494b1c62471f5f150cf034717a89376d311f895e7971 +size 938076 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/record.json new file mode 100644 index 0000000000000000000000000000000000000000..33bc4ab05a8241198f649ddab7e37b0f255f2f7f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d2_21de220e975c", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic image of a tool inside a bag, clean boundaries and sharp focus on the tool and bag", + "target_evidence": "tool inside bag", + "expected_visual_elements": [ + "tool", + "bag" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "inside", + "bag" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "21de220e975c", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.875, + "diversity_score": 0.0, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_21de220e975c", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_21de220e975c/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25827693, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 707075, + "qwen_elapsed_seconds": 5.2, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_21de220e975c/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..9f5216ce199e35a718404a7cbff2ad3a35f26e7c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ace9cfa2a22f813644ec46bbe254fa287fa998459721f4a299a951fcac16dc06 +size 526048 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..427bccfadb6a4c4b15037dfda0c38591fed48548 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:12996d5e8fbe4e76eabe337d2fcda39ef9ece66dacdfe9e14b3be0c63d59f6f9 +size 670919 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..747ea52b0457111f5ca37d0167bb07913d083a19 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b987cec3e825d9495c952c3744492e23667194dd715a3adf0d7a2fb1cff13e46 +size 605849 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/record.json new file mode 100644 index 0000000000000000000000000000000000000000..c3e650773a11c7996de139f7f8a6dc53ae2650c3 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d2_2c78cb9367b9", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic image of a bag inside a container, clean boundaries and sharp focus on the bag and container", + "target_evidence": "bag inside container", + "expected_visual_elements": [ + "bag", + "container" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "inside", + "container" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "2c78cb9367b9", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.875, + "diversity_score": 0.0, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_2c78cb9367b9", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23681224, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6345146, + "qwen_elapsed_seconds": 5.163, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_2c78cb9367b9/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..4f5351c2ea2c875b03cd918fa028f6f45fc2e89f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8accb18ac27d0113cb6453f284d0e6a68c9cf3de0e51ae0b45cb25d932b2aa49 +size 431297 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..0c931aa2ca537b9473b8439469f4af60261cc071 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1976dcbf3698e1b4a07fe0f23091647ad9684e581e578a6043c8c01484c3ed5e +size 568847 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..ba35cb0ff1cd9dae7ab3b09f8ca1a40c138bc322 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ac6a61df3d24ecfa3c69951b905905e590e1767f4128ac06a48d6aa95bb6965 +size 577924 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/record.json new file mode 100644 index 0000000000000000000000000000000000000000..26c00dc9307ef47fa9716457a40dcfc4e69d52a0 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d2_31914a9d70f8", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic setup with a box on top of a desk, clear foreground separation and high contrast between the box and desk", + "target_evidence": "box on top of desk", + "expected_visual_elements": [ + "box", + "desk" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "on top of", + "desk" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "31914a9d70f8", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.8947, + "diversity_score": 0.0, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_31914a9d70f8", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_31914a9d70f8/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23431206, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 455263, + "qwen_elapsed_seconds": 5.138, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_31914a9d70f8/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..a9580d3914422cc7b780fa6d2e4919d00a0c80fc --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a45e6ce72e9c87eb7598781a74bcbb46fdf719418a787996f83c1ec6c110f212 +size 435610 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..258b1b42dff217c8344b9af5079718fce9263ea8 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:efa3823d3efa219089e91c6a25bf53db1cf00dd0dc90dc7e50fec22bb46e4a0a +size 505589 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..68ba95e8177666a34f90ed26ac69bb34f61fd210 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b7e52ec271ee9bf9570300f7a56bd9bf87568fb313f8c4b7abe1770e5eb2f1ac +size 543473 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/record.json new file mode 100644 index 0000000000000000000000000000000000000000..d7586d480e4669eaa2f07019cafc78dd9054d6d0 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d2_40d6a9b7e720", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic image of a container with a key inside, clean boundaries and sharp focus for relation clarity", + "target_evidence": "key inside container", + "expected_visual_elements": [ + "container", + "key", + "sharp focus" + ], + "negative_prompt": "blurry, overlapping, low contrast, unrealistic lighting", + "diversity_tags": [ + "inside", + "container", + "key" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "40d6a9b7e720", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.6, + "diversity_score": 0.2, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_40d6a9b7e720", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 27511360, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6786192, + "qwen_elapsed_seconds": 5.134, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_40d6a9b7e720/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..f9b11c8e91ae0eb5ec28dcd134737c3a15851feb --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:935a959685e69bc333d91ca1d98c0368318ee8929e6eb334206a9062d1118e95 +size 357275 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..3304e904799ba461c62db9712e6c4be5c746d309 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6ae5ceab4009062978fe39cebd428befa989d7399bf69ef042e44446fc654cf +size 558491 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..1b2a881e1f3101ed7106656100109a51385b7424 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:877dc95e7955da9eee103899ee5df8212d1ec243a15ce1fdc08dca76e7060f1b +size 602440 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/record.json new file mode 100644 index 0000000000000000000000000000000000000000..f981d83e0b7bf19d4cc05980c9a0639a3fcf42bd --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d2_412c56ac4795", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic setup with a bag on top of a shelf, high contrast and clear foreground separation for visual inspection", + "target_evidence": "bag on top of shelf", + "expected_visual_elements": [ + "bag", + "shelf", + "clear foreground" + ], + "negative_prompt": "overlapping, low detail, unrealistic lighting, blurry edges", + "diversity_tags": [ + "on top of", + "shelf", + "bag" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "412c56ac4795", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.8095, + "diversity_score": 0.0, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_412c56ac4795", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_412c56ac4795/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24115452, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 414113, + "qwen_elapsed_seconds": 5.127, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_412c56ac4795/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..5d006530fda2e05adea6420fe64a1c44f54725e8 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e7463b7942eb3a2676b3631436e5c5173b4ddbe2251e6c9e8b3017c0ef9decad +size 378866 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..3c0858d9af0a68c601f8be9a110e5efb0fb60ba1 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe5847e2a7c4c5f5a1486fda8aebc69882b1403e28ef02d0e1dd02ee591c1db3 +size 567280 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..00753dd603070a0eb0f9f56fed3bc2095a804f49 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:975ca6aaf51a700e6a1026a08fc943cf7c0bcce98738fa0bb29efbc04a7e1570 +size 577912 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/record.json new file mode 100644 index 0000000000000000000000000000000000000000..7089804a21603e601416bd328b400f69aeb6bfc7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d2_4db6b2035025", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic image of a tool attached to a shelf, clear foreground separation and high contrast between the tool and shelf", + "target_evidence": "tool attached to shelf", + "expected_visual_elements": [ + "tool", + "shelf" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "attached to", + "shelf" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "4db6b2035025", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.5909, + "diversity_score": 0.2091, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_4db6b2035025", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_4db6b2035025/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20963077, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 963109, + "qwen_elapsed_seconds": 5.107, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_4db6b2035025/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..87693eab93410737f13e51d091b7587ae171b8bd --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4cd52272f8678d26892fd8d5e43fa199ff17b4c58450f1e67c409299f734c219 +size 395295 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..2bf890c09c481a43ab0074020fe5f8b094da88b1 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f57271148eb59ae71edcc6ac22624ce486d42137c03784663011146e91f1e5d5 +size 555782 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..fde69e70ca8f4b275103adc11d19c1855ee56fb7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e79185a33223b106edfab4c94181186602756ce7029a13a96d611a513aeecdb0 +size 594448 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/record.json new file mode 100644 index 0000000000000000000000000000000000000000..29a89547f2563ad961b6c4470d48e014d14a090c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d2_5126e8e36b8d", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic scene with a mannequin holding a wrench, clean boundaries and sharp focus on the tool and hand", + "target_evidence": "mannequin holding wrench", + "expected_visual_elements": [ + "mannequin", + "wrench" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "holding", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "5126e8e36b8d", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.8333, + "diversity_score": 0.0, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_5126e8e36b8d", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20810882, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 450326, + "qwen_elapsed_seconds": 5.199, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_5126e8e36b8d/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..a7926e6ab58479e1e71999e0f22a92917e844eb4 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:564082f7282028af96fd504466e68ede991a2a1305ff58c20eda8525561739bc +size 373035 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..fadf2d23c92d11abf7ee731d6a68e81fe76a709e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eeee4676a773bf6b7ba6dd164552dd0fc827275324c103820f7b383b71b42358 +size 522395 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..69cba4c41faf8a8d786207c75648042ee16b1201 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bae565721887c893186074316e7a98b5ba89fa5196f8aee1c488503dff36117e +size 523849 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/record.json new file mode 100644 index 0000000000000000000000000000000000000000..abe18591b595c29c5d18b7aa8f2d2f685d40c93b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d2_587f3e361075", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic scene with a box behind a chair, clean boundaries and sharp focus for clear relation inspection", + "target_evidence": "box behind chair", + "expected_visual_elements": [ + "box", + "chair", + "sharp focus" + ], + "negative_prompt": "blurry, overlapping, low contrast, unrealistic lighting", + "diversity_tags": [ + "behind", + "chair", + "box" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "587f3e361075", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.7143, + "diversity_score": 0.0857, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_587f3e361075", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_587f3e361075/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 27876213, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 3835553, + "qwen_elapsed_seconds": 5.224, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_587f3e361075/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..89bd2611d06a0d9c19ee861aa58d52801c1e0ee9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:11da03a44780137ae1a485bc01aebe325849def425facd588f688b0452c4af03 +size 503383 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..d56b60b6419c1a4f567789bd665f9b7587eb80c3 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f1b573793b8a605c373434befc0a0b56bd8fb5d793e99b809629087c9032871d +size 509427 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..65c5a53a3c66928f9f42fc38162618a3ac0ab38f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5027c23a6e6ab32ea02fc63faa9d95635232ab962bfbd1546139fe00b6c21015 +size 462863 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/record.json new file mode 100644 index 0000000000000000000000000000000000000000..f1911ae97e6c4b290330dc3399bc6c03770844d3 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d2_639e608a1f85", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic setup with a box on top of a shelf, clear foreground separation and high contrast between the box and shelf", + "target_evidence": "box on top of shelf", + "expected_visual_elements": [ + "box", + "shelf" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "on top of", + "shelf" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "639e608a1f85", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.8947, + "diversity_score": 0.0, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_639e608a1f85", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_639e608a1f85/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24140608, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 4550962, + "qwen_elapsed_seconds": 5.176, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_639e608a1f85/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..0c5a6e0c5af9b59a9e7d10ba5268049187d44a3a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a401f5068453e667a31bc52449b693e81b78a379662c9a8fb7ce6480f56b546c +size 187034 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..777ecd9a5cb7614f1326bc69e0767346021841c5 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:66beed319a9ef7bee82e8f01dea25763f5dda117933bd77e0bf3a42c804e8f45 +size 567870 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..d866e9be3432868e6274d18ce36ba8f180662e18 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c057ac8c54f8334fd316d411e67c2f324f7b49a49bf751302a9abddb9ce9076c +size 570865 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/record.json new file mode 100644 index 0000000000000000000000000000000000000000..65d586ee39772dbb16d10caa164b3d2d6ca763df --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d2_6ab23ebd01c6", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic image of a mannequin holding a cup, clean boundaries and sharp focus for clear relation inspection", + "target_evidence": "mannequin holding cup", + "expected_visual_elements": [ + "mannequin", + "cup", + "sharp focus" + ], + "negative_prompt": "blurry, overlapping, low contrast, unrealistic lighting", + "diversity_tags": [ + "holding", + "mannequin", + "cup" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "6ab23ebd01c6", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.6364, + "diversity_score": 0.1636, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_6ab23ebd01c6", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 22240429, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9874851, + "qwen_elapsed_seconds": 5.129, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_6ab23ebd01c6/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..9e52ee814ce77df4effe833a7137884ab82e1d50 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:daeab2a4bae149c85f35cde1332b10a15665d0d3de4f778a054ec01598ee8bc6 +size 567505 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..21ccd92724147b48f6514e143bf0217eba11df07 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:498498fecd40ad69361051c746f41271b9c4f357803ed73d30dd62755ac7e1d5 +size 608089 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..5738a4d86bb8091500fc2333da85f4f510535962 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5dc7c4553315a5ab908380cae4b5f6416f209f0dd4b04fa267112d6b0d4d9beb +size 699284 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/record.json new file mode 100644 index 0000000000000000000000000000000000000000..28e639adfeedfe2e373f84554ed78597405de273 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d2_6b2da3ae773a", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic scene with a bag inside a box, clear foreground separation and high contrast for visual inspection", + "target_evidence": "bag inside box", + "expected_visual_elements": [ + "bag", + "box", + "clear foreground" + ], + "negative_prompt": "overlapping, low detail, unrealistic lighting, blurry edges", + "diversity_tags": [ + "inside", + "box", + "bag" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "6b2da3ae773a", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.6667, + "diversity_score": 0.1333, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_6b2da3ae773a", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 30250693, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9963444, + "qwen_elapsed_seconds": 5.202, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_6b2da3ae773a/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..fd70cfcd5b48f75b6ced19fe2b2420aa1ac3b130 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f15126efadb5b7de86d2f0eca817922158b53b81174bced12f95c4449ecd6ff5 +size 359032 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..f39927d777a248fd13419469f302fe593e3a0180 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be6bb8f37056ea607538c220d5e43c0d1c6110f0cab1dd5e4e66255890b732bb +size 398803 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..b62ef5b85f20ee1d685fadbcf1f7a9e03082b53f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:378de3828de7bf70da9bf2130db78dea1be84d9e170d067721f0764b5f16b6d5 +size 563956 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/record.json new file mode 100644 index 0000000000000000000000000000000000000000..ad3d2edb833e819edb6d2361bbbf595c5c22696a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d2_6bbeb1d0bad7", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic setup with a tool under a table, clean boundaries and sharp focus on the tool and table", + "target_evidence": "tool under table", + "expected_visual_elements": [ + "tool", + "table" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "under", + "table" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "6bbeb1d0bad7", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.619, + "diversity_score": 0.181, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_6bbeb1d0bad7", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26374483, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 3150863, + "qwen_elapsed_seconds": 5.181, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_6bbeb1d0bad7/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..d0ecc33c80dda9e3e5b43fc3dd55d4ef348ce1f5 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0bb2d2e0ebb66952ad1d6bd9a30d1c283a3c78aace7073b4de319488d431f220 +size 363953 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..38511225860d738e24e2895684de6d27b5017596 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:186fe7449ea347e970c7f0ed0bad58c3ab333dee1d76097d0d827f45cdeda743 +size 557052 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..b5a8ba7a8bcdda4e943b19df31e95affd29dc7cf --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f3fdb0b48fc1d379acb9965b7bb415e26d951fca61b36b2e1af03fad448d1490 +size 528419 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/record.json new file mode 100644 index 0000000000000000000000000000000000000000..5c6fb7b6660ceebe9c6f6a006937d95d3262ede4 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d2_81fb8f2bf58d", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic scene with a mannequin holding a sign, clear foreground separation and high contrast between the hand and sign", + "target_evidence": "mannequin holding sign", + "expected_visual_elements": [ + "mannequin", + "sign" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "holding", + "sign" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "81fb8f2bf58d", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.7895, + "diversity_score": 0.0105, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_81fb8f2bf58d", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20501665, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 4851240, + "qwen_elapsed_seconds": 5.19, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_81fb8f2bf58d/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..c677f1349ef59eecabf06184bb11b2ca17896783 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4eb046d392cf4e58d6c1baf76a35199abd646b9cdc8f96cf5260bc8e152d7449 +size 227530 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..aa97f8733c8f9b6a879212530e4eb9886c7fd681 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e5259627d8ec148a187f6ddf15b9d8b59977a0600b3456754052835d742276d +size 596914 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..0290970caada81e97c70c1933d59b89451b7d7ac --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1182bb240b4b89536e6b4655807eba19568325f9da04e7ba6c4753d773e16e8e +size 654598 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/record.json new file mode 100644 index 0000000000000000000000000000000000000000..18355f9d1c5b2e71ba8d862515e24d387eb13d57 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d2_add4beb155f4", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic image of a mannequin with a hat on its head, high contrast and clear foreground separation for visual inspection", + "target_evidence": "hat on mannequin head", + "expected_visual_elements": [ + "mannequin", + "hat", + "clear foreground" + ], + "negative_prompt": "overlapping, low detail, unrealistic lighting, blurry edges", + "diversity_tags": [ + "on top of", + "mannequin", + "hat" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "add4beb155f4", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.625, + "diversity_score": 0.175, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_add4beb155f4", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_add4beb155f4/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29113177, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2090035, + "qwen_elapsed_seconds": 5.179, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_add4beb155f4/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..f0e9b024623d039aa01693931424ab8bfebb075a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f1b31081a0e43f668db13bb5d61f714a4e1d6ce4704024fd6201f7ebf1f0082 +size 484790 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..98a1766d0c1f7bf7f471ba42605e042f15fcd335 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fafe6ab6cabe3fed6ace8fba972c32ba2bc95f766bb9e28f7200fbe2d70110fa +size 554168 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..bca34648b41e90b4a9764c32e884f3da3a3ed639 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ec8d08bc1c024e59fecc3fe4b1929647a01b714cd503388d867aaf5b1e90888 +size 525262 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/record.json new file mode 100644 index 0000000000000000000000000000000000000000..6bd7bd9cc6160d274e9d2fa0aa79799d6aeb4cd3 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d2_b1a4e86c9249", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic scene with a sign behind a mannequin, clear foreground separation and high contrast between the sign and figure", + "target_evidence": "sign behind mannequin", + "expected_visual_elements": [ + "sign", + "mannequin" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "behind", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "b1a4e86c9249", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.7895, + "diversity_score": 0.0105, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_b1a4e86c9249", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24435050, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1438495, + "qwen_elapsed_seconds": 5.222, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_b1a4e86c9249/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..3891b187c2d2e3d96d15887dd13b4a08c28db4a7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:62cbfe9c33525e9032f79c7b9b48a2ddaefed87807cfc49c62b8a45dda33320b +size 571569 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..e4a99be2c129afd3955fee1155036f260fb92127 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:855cccdc27849161ba49c89f1661d28df6493859260b058dbd200353e07bb9ad +size 487381 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..075f27a66fbaade77e22e6d75a25ee4c471a20cf --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:25d039d36ad22679b7cff89eea452aa981176d213f392681035135eac4543e7d +size 494304 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/record.json new file mode 100644 index 0000000000000000000000000000000000000000..811e248bb4b41ac1c050414c37c0b1080bd6672e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d2_dd6c090b67d8", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic setup with a book under a lamp, clean boundaries and sharp focus for clear relation inspection", + "target_evidence": "book under lamp", + "expected_visual_elements": [ + "book", + "lamp", + "sharp focus" + ], + "negative_prompt": "blurry, overlapping, low contrast, unrealistic lighting", + "diversity_tags": [ + "under", + "lamp", + "book" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "dd6c090b67d8", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.619, + "diversity_score": 0.181, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_dd6c090b67d8", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24632545, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 4669038, + "qwen_elapsed_seconds": 5.189, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_dd6c090b67d8/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..0f6cff6de9cd5ad779139598348cddead4e3cc1d --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9410e370b49932b026e8cd8cb748cea66b0f928654404b3c684bdc49e577a11 +size 570147 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..d26ea605c7015ac90d8d1e3b5014858a22155225 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:afb0199968797d5cfe865f375eac73aefe77ffe59eb507598725c717c7dd3ff1 +size 607970 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..501f8251a1c1b210d97b722c4f8fff64147ae51e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:083ecc165ea7b1b017c71ad3b8eeda5160b5cc5aa082b6ee8830df66c7dcfe08 +size 595547 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/record.json new file mode 100644 index 0000000000000000000000000000000000000000..2384448734ba6da3274ee775d4497c5f90796ce7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d2_e10e5f2a5b78", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic scene with a mannequin holding a wrench above a toolbox, clean boundaries and sharp focus for clear relation inspection", + "target_evidence": "mannequin holding wrench", + "expected_visual_elements": [ + "mannequin", + "wrench", + "toolbox" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", + "diversity_tags": [ + "holding", + "tools", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "e10e5f2a5b78", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.7273, + "diversity_score": 0.0727, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_e10e5f2a5b78", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20484070, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1626980, + "qwen_elapsed_seconds": 5.168, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_e10e5f2a5b78/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..3e4a0806ab2c5eba142fd952c3ce9842a56ddd7f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08513bda6ec1ee7915a20ad93ee9da8579c1a69bac631f796c5c570a84d52f04 +size 407510 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..001d151aae20a2a5563f6915e17cd6a298517995 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d0b11903108f54a1721257d6307b0b74a30980b0f5217a58e11988bfe76aef7a +size 480571 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..d4e4877eb4367550292d25fd04f16b6923b0cb97 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c0e97936019e9a5437d970028ae1e3eb8f4f6d22a9e6c08835e985e423022d5 +size 581029 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/record.json new file mode 100644 index 0000000000000000000000000000000000000000..b23dd6394cc823aa3286a0ffebfd4f107dd9772c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d2_f48653185696", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 2, + "image_prompt": "synthetic photorealistic setup with a tool on top of a desk, high contrast and clear foreground separation for visual inspection", + "target_evidence": "tool on desk", + "expected_visual_elements": [ + "tool", + "desk", + "clear foreground" + ], + "negative_prompt": "overlapping, low detail, unrealistic lighting, blurry edges", + "diversity_tags": [ + "on top of", + "desk", + "tool" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "f48653185696", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.8095, + "diversity_score": 0.0, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d2_f48653185696", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d2_f48653185696/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29235481, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5328890, + "qwen_elapsed_seconds": 5.153, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d2_f48653185696/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..dce7dbb72318e168bb733f18efabb392535e2dd9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5913b7ce09b13cc0626a1f8df65a4634c378f466df16c4b39dca639029ba5615 +size 433737 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..d20a37c9ff798faf3eaeb22522fe9b9caa5c57a7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5624847796b7655ccbbbfe0124757c2a40fa9b4d8995586e38714e38b9732178 +size 559015 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..4f156ae2f98d04f5f359b37860ce9457b1d8d3ef --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2d56522e26fd6bf93a2f556201198b6e42a289f7dee76b92d7980ee0ebfe9f8 +size 612121 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/record.json new file mode 100644 index 0000000000000000000000000000000000000000..d081464751aee2bc57374a98f9d7dc2708fa4b48 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d3_074db631b159", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic setup with a bag on a desk, a pen inside the bag, and a sign next to the desk, high contrast and clear separation.", + "target_evidence": "pen inside bag", + "expected_visual_elements": [ + "bag", + "pen", + "desk", + "sign" + ], + "negative_prompt": "low resolution, unclear boundaries, cluttered background", + "diversity_tags": [ + "inside", + "desk", + "sign" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "074db631b159", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.6957, + "diversity_score": 0.1043, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_074db631b159", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_074db631b159/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25893801, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6993162, + "qwen_elapsed_seconds": 5.114, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_074db631b159/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..9147893eb15c71db06d7abb20e14d57854cbea0f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:269ba5294e5d7e1cdbb631d14964b7ec7192ac34056d2f04a9d95c16ebdbd759 +size 513123 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..10fbf12c13e4260ac7c6f4828af1c0995af8406e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:70bf9c583b0ad6290bf74228e819aa364be2efa4db91df0741f1380ea7b25c08 +size 545212 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..5a12604415e8d5111c2736dd8aad276bcd3070a9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c3bb3591f5c6402b93a5d11c3e427cff2e1376323b5d3f2b52e11bc88ee64bf +size 602395 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/record.json new file mode 100644 index 0000000000000000000000000000000000000000..4f3adc626cee7ceb5b400410b24a7f7e10ba2d05 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d3_07561e275c5b", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "synthetic photorealistic image of a tool placed inside a container on a table, sharp focus to distinguish nested objects", + "target_evidence": "tool inside container", + "expected_visual_elements": [ + "tool", + "container", + "table", + "lid" + ], + "negative_prompt": "blurry, low contrast, unclear boundaries", + "diversity_tags": [ + "inside", + "container", + "table" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "07561e275c5b", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.5909, + "diversity_score": 0.2091, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_07561e275c5b", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_07561e275c5b/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 21728305, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5979448, + "qwen_elapsed_seconds": 5.139, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_07561e275c5b/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..2315d5a9567acfe230243c93632322ea379eaa3d --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:540fc49aef91d8bd5ac63c291ba031cd7691ef9ff7eea0345e402a2a91dd2a31 +size 459243 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..5a1b00367835edbd323970d8d0c6cd2a3faf6cc2 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74945c36257af863602f468e246459d927a92c5aba08f82eef8d724d99f68bfd +size 552903 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..bfd3891dc289c83cc544497f15ec5ceedcb0ff99 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:533df0394c611d23dc6b0d6be9c2a6770aec3a4280b9a82de2aa699f86e62386 +size 588372 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/record.json new file mode 100644 index 0000000000000000000000000000000000000000..74b543d98995b27ed706c5b05cb0649a8bb88176 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d3_269228cb8c34", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "synthetic photorealistic setup with a box on top of a shelf, and a lamp behind the box, clear foreground separation", + "target_evidence": "box on shelf", + "expected_visual_elements": [ + "box", + "shelf", + "lamp", + "background" + ], + "negative_prompt": "overlapping, low detail, weak lighting", + "diversity_tags": [ + "on top of", + "shelf", + "lamp" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "269228cb8c34", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.75, + "diversity_score": 0.05, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_269228cb8c34", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_269228cb8c34/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26245538, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5381559, + "qwen_elapsed_seconds": 5.172, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_269228cb8c34/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..791671992f78deb191dcc78dfbec065259d36c3b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5eb787c08d001ac8f54244a33bfab5e9695fba7f080b3c705b5096d1005c863f +size 319240 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..d43480180c5377c9db4f298027c2dfc2f3780226 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:255e5cbe6859a3d97da2048e44aed3c5f57cedddab9629d58747fe6360381113 +size 503939 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..7260426c4b03aa8cc00b81d47bdc496928b0bf10 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc02e2f339bbdbd783e751118982bdcf0336f60d29ad826f4f84e5e2c64185fd +size 615698 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/record.json new file mode 100644 index 0000000000000000000000000000000000000000..577b7430a96ed41b596ad733b526032fe13e6d23 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d3_2a425cc15ef7", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "synthetic photorealistic view of a mannequin with a bag on its shoulder, and a box on top of the bag, clear visual separation", + "target_evidence": "bag on shoulder", + "expected_visual_elements": [ + "mannequin", + "bag", + "box", + "shoulder" + ], + "negative_prompt": "overlapping, low resolution, weak lighting", + "diversity_tags": [ + "on top of", + "bag", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "2a425cc15ef7", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic view of a", + "max_similarity_to_other_prompt": 0.5909, + "diversity_score": 0.2091, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_2a425cc15ef7", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24417931, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 8199010, + "qwen_elapsed_seconds": 5.15, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_2a425cc15ef7/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..3b0779b2bce7093da65ae8151876e62b3a2514c5 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7d301e5416b71ef08c0febb44185e8baa293ddc2f00854c80e7c6220cff0ab5 +size 442898 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..112b18d29af0ed8e869a4596368f2c7fa03060d9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9640de34cdca134904721660cb52045d70808538fcab09fc97a41a083d001093 +size 585483 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..4b2dfae73b07a194a3c406625681202abd93c112 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3e8e1cb23b505ebf112796e6f831af7ffb8c507cf541327eecefcc47191230ab +size 611897 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/record.json new file mode 100644 index 0000000000000000000000000000000000000000..6415f908723598c024feecac666136ffcfe66b97 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d3_3e33c1f92393", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "synthetic photorealistic setup with a mannequin sitting on a stool, and a bag on top of the stool, sharp focus for clear relation", + "target_evidence": "bag on stool", + "expected_visual_elements": [ + "mannequin", + "stool", + "bag", + "seat" + ], + "negative_prompt": "blurry, low contrast, unclear positioning", + "diversity_tags": [ + "on top of", + "stool", + "bag" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "3e33c1f92393", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.52, + "diversity_score": 0.28, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_3e33c1f92393", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_3e33c1f92393/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20705994, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1618873, + "qwen_elapsed_seconds": 5.169, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_3e33c1f92393/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..87b85145e2672a16ba15edf5c35530a377a8c35e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e88b50d4fed9e832258d6ecd7f0ea96a76349958bf3864f645873bd6f8ed4ea7 +size 433298 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..6f43393bdb597ae4eb56e65344f7531c14816c25 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd89f43dc7b89885dfd85f0d598720836033a205a40bd5484dd1ce14a6c0d0dd +size 605719 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..92cd3629ed5348805c513e90174b81124aa30480 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:66e07e2da403220f2607ebb3f4d9497220dbb9a8b21a3f2537d892b93690d578 +size 542845 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/record.json new file mode 100644 index 0000000000000000000000000000000000000000..94a3648fcae1958478d120409d014e72ae6bc30a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d3_427f77eb50cf", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "synthetic photorealistic setup with a bag inside a container on a table, sharp focus to distinguish the nested objects clearly", + "target_evidence": "bag inside container", + "expected_visual_elements": [ + "bag", + "container", + "table", + "container lid" + ], + "negative_prompt": "overlapping, low resolution, unclear boundaries", + "diversity_tags": [ + "inside", + "container", + "table" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "427f77eb50cf", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.5909, + "diversity_score": 0.2091, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_427f77eb50cf", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_427f77eb50cf/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24940400, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9012894, + "qwen_elapsed_seconds": 5.113, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_427f77eb50cf/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..d7a84172d2b434655f4a63e2d869e3268dbf0829 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c7b9d6c8570460e1c35a45448dc15eaeaa2c2c56e6076f5e7441e0f66895093f +size 431941 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..082be6e063a0d267287a6e45f3905b998f0231c9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:429af389f150dc5606be8908589ddcd1de89842515b301285d5a4325ddff6fe0 +size 539626 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..2edf1ad4e290d6079f3e93ebf705c1cf00c34fa9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d7514d1248ebe9a7bcf6c23fb3f74ab3677b7ae53a401fcb9fc08dc643409ea +size 525203 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/record.json new file mode 100644 index 0000000000000000000000000000000000000000..0fab0acffd9cae49a54d781ad1bb3ef336f5ad1f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d3_4356499ba7ce", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic view of a mannequin holding a sign with a tool, the sign and tool clearly visible with clean boundaries.", + "target_evidence": "mannequin holding sign with tool", + "expected_visual_elements": [ + "mannequin", + "sign", + "tool" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", + "diversity_tags": [ + "holding", + "sign", + "tool" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "4356499ba7ce", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic view of a", + "max_similarity_to_other_prompt": 0.5455, + "diversity_score": 0.2545, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_4356499ba7ce", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_4356499ba7ce/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 27164474, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2502063, + "qwen_elapsed_seconds": 5.07, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_4356499ba7ce/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..320837820c4cb9d953188116ea2be36a0e5f60e1 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd0d1cdfba75755edbb42691941ae5048d4c37c76be8e8a8153d76bb026c8c7d +size 61722 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..18efa2fbb678b58eb13595deae0fe3a4df7e6fa1 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0df2bf349c8ea7f8db8b9a6a392a5d4301106bade8183c4f78e59a47170a0ec7 +size 620733 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..5a60654ec988d37acab27b218f2481cbbf7639a7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ddfc842eb7f22ac74285f21be299fdcd931dcb5f41b6e993391387f39b07b117 +size 543264 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/record.json new file mode 100644 index 0000000000000000000000000000000000000000..50a413beec73140456e71efaa6b5f7770d438374 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d3_4538c668915e", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic setup with a tool inside a box under a shelf, the tool clearly visible with high contrast.", + "target_evidence": "tool inside box under shelf", + "expected_visual_elements": [ + "tool", + "box", + "shelf" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", + "diversity_tags": [ + "inside", + "under", + "multi-step" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "4538c668915e", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.6667, + "diversity_score": 0.1333, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_4538c668915e", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_4538c668915e/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29243261, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 986467, + "qwen_elapsed_seconds": 5.17, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_4538c668915e/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..ea6a0edb6216feb02a27ecc5bed704641559b426 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a55e42d1aa2c9f847e063bbe1441497d73ce346c44c97cb2690d906045c9caf3 +size 443155 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..d3553128647af9ed2889baead2e9fbfdb25c3246 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:59721f911cd94ebe785a1599f4394c5ac2fd3816b031c6421dae00599d60ce17 +size 500397 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..11645b43f14881e2f0dd8fb01eb5886110310b41 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77f7dfe874b030d0353a54d04ca55f9c112dd1932616deed772b008ecebc0416 +size 515696 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/record.json new file mode 100644 index 0000000000000000000000000000000000000000..e04b4d7eec3883af7308cfa47ae8b538999b2904 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d3_5555f1e89a14", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic setup with a box on top of a shelf, the box clearly visible and separated from the shelf edges.", + "target_evidence": "box on shelf", + "expected_visual_elements": [ + "box", + "shelf" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", + "diversity_tags": [ + "on top of", + "shelf" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "5555f1e89a14", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.6, + "diversity_score": 0.2, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_5555f1e89a14", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_5555f1e89a14/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20727461, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6278145, + "qwen_elapsed_seconds": 5.214, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_5555f1e89a14/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..c218738aa2d80350ddd16eb2afbf790c83855eb9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f6fde61256a3a47a1b240b3262173450dad63098e283a1395790b295d425988 +size 73109 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..bf078bf8622fbef96bed765d47bb063601e32a93 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b0f17f2452cacc495a7c3ce2a3833065e81c897e3c17f3602f25b3df39ce94fb +size 497697 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..6ea48b69e2bfd9e5ec654f9fbc319bfe56b4151b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:43434e50fad080ec1b8bfdeb480ddfd73ce7b239ce2ac88d961472c0600011a4 +size 548017 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/record.json new file mode 100644 index 0000000000000000000000000000000000000000..4eb2523296709983273d0b9fe434d4571474bef6 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d3_571cf3225cf2", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic scene with a book under a table, the book clearly visible and separated from the table surface.", + "target_evidence": "book under table", + "expected_visual_elements": [ + "book", + "table" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", + "diversity_tags": [ + "under", + "table" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "571cf3225cf2", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.4545, + "diversity_score": 0.3455, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_571cf3225cf2", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_571cf3225cf2/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24030301, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9123441, + "qwen_elapsed_seconds": 5.195, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_571cf3225cf2/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..40638fcbd2e90144b1d3d62ca5582683246a7d62 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f76796e7e51738389aca20c9e872ac2bbd3016915ddf77d2953f59efefccd7c3 +size 383802 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..ac1e193146c8bff0b43e92668a24ab76a3aac21c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7bf260e2ac6a8b3785d77f92fcc85be06774c9c46275ded6b3543eb2431af3fc +size 589111 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..44b7191d3457416cb975ff459455666236f1f476 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0eb4646ea4f9ae9cf4d800cf930cfad2b1a985d64596d5aed5c0e4c90f6b6a9d +size 562592 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/record.json new file mode 100644 index 0000000000000000000000000000000000000000..0058a100fe4c1cc525f78003611c92c0c8cd76e1 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d3_6a8fa33182a5", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "synthetic photorealistic view of a mannequin holding a key, with a box placed on top of a wooden shelf, clean boundaries for clear relation identification", + "target_evidence": "mannequin holding key", + "expected_visual_elements": [ + "mannequin", + "key", + "wooden shelf", + "box" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", + "diversity_tags": [ + "holding", + "shelf", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "6a8fa33182a5", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic view of a", + "max_similarity_to_other_prompt": 0.5385, + "diversity_score": 0.2615, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_6a8fa33182a5", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 22513879, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 572812, + "qwen_elapsed_seconds": 5.207, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_6a8fa33182a5/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..b4252a82657b921ac2d09ea6899a02bb6a0343b5 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba3165fe0e26da238df39ceb5d361f2278c39d844da5ff5887cb4b3b8382bf4d +size 373402 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..57c1e94bb0b1dcc7afe44a30628fdd522f64cfca --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7683d62a56efd6df4d33a176230a74350df42512aaf656360a0b50e7ec89cf6b +size 617119 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..ef7d3a6fe8398334e35d821bd817252d557e5df2 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bca741fa6ea8330ed04deff89cc1814350c46f5efd968a13894df9475186147e +size 554549 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/record.json new file mode 100644 index 0000000000000000000000000000000000000000..fcc94f5ea16aeede48a1ac492e0f97de0c5b718d --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d3_810283167dc6", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "synthetic photorealistic scene with a mannequin holding a sign, and a tool attached to the sign, clean boundaries for clear relation", + "target_evidence": "mannequin holding sign", + "expected_visual_elements": [ + "mannequin", + "sign", + "tool", + "attachment point" + ], + "negative_prompt": "blurry, low contrast, poor lighting", + "diversity_tags": [ + "holding", + "sign", + "tool" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "810283167dc6", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.65, + "diversity_score": 0.15, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_810283167dc6", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_810283167dc6/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23566118, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7692250, + "qwen_elapsed_seconds": 5.147, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_810283167dc6/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..e9890a9dfaa263617f25eb905839affa0d2d3cc0 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac0e93835f592e105f0f045de3d76e2869372ae3a46d36c39ec29fd29e2d5d49 +size 433526 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..259b9b5fef1e56705230e8f5d7cc1ac754eb0181 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c057211125188d25f526fffe20b66c534ed1596b6a3f0070ae29ed7a0192a5d9 +size 541892 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..1946002390f6547194f54e33b101f54884b19479 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a9fdae587bcae52d266302eec60e7460a0d88f30c8856bf018ed2068afdd0f4d +size 589672 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/record.json new file mode 100644 index 0000000000000000000000000000000000000000..b1a4c23b7bdae831c40d1e213db05aafc0f7cf9d --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d3_8ccffd73bf0e", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic setup with a tool on top of a box, the tool clearly visible with high contrast.", + "target_evidence": "tool on box", + "expected_visual_elements": [ + "tool", + "box" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", + "diversity_tags": [ + "on top of", + "box" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "8ccffd73bf0e", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.6667, + "diversity_score": 0.1333, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_8ccffd73bf0e", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28326691, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 3831477, + "qwen_elapsed_seconds": 5.107, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_8ccffd73bf0e/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..006b341fc32de8eca8468a917166bff5b8d5be40 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e40b5a45acf0216353070eeadc62d80af8d151018ffd5c7cc71871eaf3ab6c14 +size 189065 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..97fe19aab1e96c0d7129327c57a78de1f14b0ae7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c61742eaaa6f6b7dbc848fa2a65b16e8f8ef09a6208f64304d6fe1c1cee845a7 +size 419392 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..78663b6dbda785e6d07a92a75637971903bcadec --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9f17491580d34f305e09010438cc1b718cfa89bbe32f9567577ee5aed722cd19 +size 494837 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/record.json new file mode 100644 index 0000000000000000000000000000000000000000..f04e6d72f5384c2fe79e43f424f61a565b998d17 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d3_cc3de276f8b9", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic view of a mannequin standing on a box, with a container on top of the box, clean foreground and sharp details.", + "target_evidence": "mannequin standing on box", + "expected_visual_elements": [ + "mannequin", + "box", + "container" + ], + "negative_prompt": "overlapping objects, unrealistic colors, low detail", + "diversity_tags": [ + "standing on", + "box", + "container" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "cc3de276f8b9", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic view of a", + "max_similarity_to_other_prompt": 0.52, + "diversity_score": 0.28, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_cc3de276f8b9", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23327583, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5783738, + "qwen_elapsed_seconds": 5.164, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_cc3de276f8b9/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..8882f6cd6c287708942763deb582367f727d518e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2e729c79e7bbb8324bb5173aed5b6a04b81dc025cdbccc65a235aba65ab0e34 +size 430432 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..47ab5341d482f3441d6b99e64fe4929865812b14 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d12b4c6bde2189ed75ebec1bd37bcfe60212b12ec32d8b67b986d4eb4a5311c +size 553168 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..337faed0f87d427f8d08069a8544791f8a7f2e60 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18becf9c80ba1d3a9c0c1c2d06c5495b930364006208f42ab1db3615b726d49e +size 633803 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/record.json new file mode 100644 index 0000000000000000000000000000000000000000..70c28a25e211a76ec6e9f68a92fe82855d0d506b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d3_cdfc1eaab7f0", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "synthetic photorealistic image of a tool attached to a sign, with a clear line of sight between the objects for relation clarity", + "target_evidence": "tool attached to sign", + "expected_visual_elements": [ + "tool", + "sign", + "metal pole", + "attachment point" + ], + "negative_prompt": "ambiguous positioning, poor lighting, blurry edges", + "diversity_tags": [ + "attached-to", + "sign", + "tool" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "cdfc1eaab7f0", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.5217, + "diversity_score": 0.2783, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_cdfc1eaab7f0", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28200127, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2247326, + "qwen_elapsed_seconds": 5.11, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_cdfc1eaab7f0/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..5c138db153be8fee9153c89b08b4d21b87be90c9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af8cef82331cb3ebbd707d799dab30a8c7e9746710411bde4a51216fc1c4e990 +size 508133 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..fef825ecd4573f5e996acfb351b81ece23803689 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:03c17f8486d28ad3852411bdd74de0e8f245adb44437abd37500ae432898aee6 +size 621986 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..68aafdc3d32db8bb2aee8146eb601e3ca9a60729 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:88e77b6b70895f34b9b33b96c6a6a1c4c5f32ffadd517ac78d75041b35dda9d5 +size 569444 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/record.json new file mode 100644 index 0000000000000000000000000000000000000000..7e849a3d0879ba9590a16944eb1687ee93f0b073 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d3_db89b92548af", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic scene with a mannequin holding a tool, placed on a shelf with a box behind it, clean boundaries and sharp focus.", + "target_evidence": "mannequin holding tool", + "expected_visual_elements": [ + "mannequin", + "tool", + "shelf", + "box" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", + "diversity_tags": [ + "holding", + "shelf", + "box" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "db89b92548af", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.6522, + "diversity_score": 0.1478, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_db89b92548af", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_db89b92548af/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28352076, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2392804, + "qwen_elapsed_seconds": 5.219, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_db89b92548af/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..12ef79440680a0820591f0cbc6027c57430624ac --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e7929d9fc3f144ef960bcff6116123495c44fba89faaeb791454e6b84a0f7d87 +size 524944 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..8295871d0439372eef2377732959b03d213eeb84 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:14f5a173c0472508eafdddd100e24577377051d2e123f08cfa8ac9d2e6f99c3b +size 615169 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..b84a543f5f209a57eef45f5594c803ea1a826cdc --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:805470c85fa37bbf638197bb3bfaac84739220fca467705f8a8295ba605f3d75 +size 604298 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/record.json new file mode 100644 index 0000000000000000000000000000000000000000..4f33d0a4ba3034b8ffe9e1f68072a2d7a3c6d3fe --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d3_dc946d1ff598", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic image of a mannequin with a bag on its shoulder, the bag clearly visible with clean separation.", + "target_evidence": "mannequin with bag on shoulder", + "expected_visual_elements": [ + "mannequin", + "bag" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", + "diversity_tags": [ + "on top of", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "dc946d1ff598", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.5455, + "diversity_score": 0.2545, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_dc946d1ff598", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_dc946d1ff598/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23215044, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1209501, + "qwen_elapsed_seconds": 5.144, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_dc946d1ff598/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..b6154be4a795581bf3a28d03b8b1e9a780926389 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e930dd820afb40de687f4526f0d5118511f00e342c42dd8770a99d83566df057 +size 413598 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..8a662c4672dcb36c0f95e0859e5730d0fac22f0f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f99e9b2d8593c91c81b49ba3c690417fa1a8b793b5e5404bb6310846c4615fc3 +size 603994 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..2e9e369f4ede9a9ba01946a1fa91e933896196a2 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9266579a31fc7fa046a6b8eadfc9eac617c345cda759e62d13d089676b694f07 +size 601263 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/record.json new file mode 100644 index 0000000000000000000000000000000000000000..92d66c339d9243727d065f1865d1c08d0a30856e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d3_e0705ad77721", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic scene with a mannequin holding a key, the key clearly visible in front of the mannequin's hand.", + "target_evidence": "mannequin holding key", + "expected_visual_elements": [ + "mannequin", + "key" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, poor lighting", + "diversity_tags": [ + "holding", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "e0705ad77721", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.5217, + "diversity_score": 0.2783, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_e0705ad77721", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_e0705ad77721/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28312520, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 567252, + "qwen_elapsed_seconds": 5.227, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_e0705ad77721/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..71721c87e8751368c994ff2aa315265cf360671a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44c435abad00f2511e84121dc47cba887c96e3694f1bc974b392b6dd7520ce61 +size 499594 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..5de3fbe0193ee0752bd92a37965d35f3b87829dd --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a6423d9deb0dd24ff7a406b74dc4fc77778a6a64611e658482dd584cd6984d61 +size 590732 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..774ce74a02edf1508fde44e9d3969b93a826e2aa --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:90ac31543c9d54c645a82ea513a017751d529fa917fb05ce272e25f0502b9940 +size 581474 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/record.json new file mode 100644 index 0000000000000000000000000000000000000000..4cf53ac01c7469ee1a1edae926dcd2fdd27f7624 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d3_e2e5bfbc9375", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic image with a tool attached to a tool rack, placed under a sign, clear foreground and high contrast.", + "target_evidence": "tool attached to rack", + "expected_visual_elements": [ + "tool", + "tool rack", + "sign" + ], + "negative_prompt": "blurred edges, low lighting, unrealistic textures", + "diversity_tags": [ + "attached to", + "tool rack", + "sign" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "e2e5bfbc9375", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image with a", + "max_similarity_to_other_prompt": 0.5455, + "diversity_score": 0.2545, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_e2e5bfbc9375", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29470029, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2084011, + "qwen_elapsed_seconds": 5.228, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_e2e5bfbc9375/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..319734189aa759b39fbe04e55861841a96161f83 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af57dc11a7d0b381acd5ba8f202958140ecaf6951ec5a6e2e31e4faf7c89b21d +size 373399 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..c23662a8ef304857e6662370e2056d25d4634aea --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5007e60c3fffb5f5e3cb54ab5562f10671b73ca25878c52788f3b6d94fc2e17d +size 600839 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..9bf210ffe45cb1361cea0bf602f63e759fbbdd07 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:04f3c9718ef19aaeb08354740e0ba9ea339c21d8ea38ada3cb82d875ffa1365b +size 586256 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/record.json new file mode 100644 index 0000000000000000000000000000000000000000..143a04f55f16c333f24ab11cee6a90285ec7e5b7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d3_f711a1758ab2", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 3, + "image_prompt": "Synthetic photorealistic scene with a mannequin holding a box, which contains a key, clear boundaries and sharp focus.", + "target_evidence": "mannequin holding box containing key", + "expected_visual_elements": [ + "mannequin", + "box", + "key" + ], + "negative_prompt": "overlapping elements, low resolution, unclear objects", + "diversity_tags": [ + "holding", + "box", + "key" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "f711a1758ab2", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.6, + "diversity_score": 0.2, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d3_f711a1758ab2", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d3_f711a1758ab2/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20526890, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 979867, + "qwen_elapsed_seconds": 5.177, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d3_f711a1758ab2/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..898e2d57950f75680fb4817bdb07cf1974b5022d --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:029340bb2bd9d0c406657b75c21a733be57d3f41d159b8f1bec3b59b21210e66 +size 500182 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..5e26c0d6c1086718d192cd8286bb32455b482de7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9b9879bf1e97bb3fe62d30bf7bbd353a3a3d9d8e41995bc618073bab30cb89c1 +size 636529 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..b91fa86ddd7abf441100505e8f06cdf14969847f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c49e08087fbf27c6523e4556a3920a217bcada5a652cbe8d255228be5531f26 +size 586759 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/record.json new file mode 100644 index 0000000000000000000000000000000000000000..c047522dd6bd7c46631f45974a2be379c7fdb744 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d4_051ad7519d81", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic setup with a bag on a desk, a pen inside the bag, and a notebook on top of the desk, high contrast, clear foreground separation", + "target_evidence": "pen inside bag", + "expected_visual_elements": [ + "bag", + "pen", + "notebook", + "desk" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", + "diversity_tags": [ + "inside", + "bag", + "desk" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "051ad7519d81", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.7391, + "diversity_score": 0.0609, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_051ad7519d81", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_051ad7519d81/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20416768, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7510431, + "qwen_elapsed_seconds": 5.209, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_051ad7519d81/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..45c8e63b0b6b5c794277f843466225f6888ec5b0 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44f6cb8c7dabcb28a1aa4e7e2802fb56c14afe776cb086833b39e4c316847b00 +size 417701 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..a8b62d242ecd92ed8e1e745bfa7eff720aa87f2e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8aed8ed4c78be398fbd0d2e2ac8c326d5e582ab3d8953ccdd58946ad924bd2e0 +size 582679 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..bdefabf17bae7c7fde3586e0d544a0832ad3eb97 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:712abce11281fdf62ed23ee5f21ca5170fef3af349a2247f8b0edeea03bae3a0 +size 577711 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/record.json new file mode 100644 index 0000000000000000000000000000000000000000..80d30521ec98effa576d69f304ab58f9a6cdab01 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d4_0d4dcd9caae2", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic image with a tool attached to a sign, sharp focus and clean edges for visual legibility of the relation", + "target_evidence": "tool attached to sign", + "expected_visual_elements": [ + "tool", + "sign", + "wall" + ], + "negative_prompt": "low resolution, background distractions, soft focus", + "diversity_tags": [ + "attached-to", + "sign", + "tool" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "0d4dcd9caae2", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image with a", + "max_similarity_to_other_prompt": 0.52, + "diversity_score": 0.28, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_0d4dcd9caae2", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28008680, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9649498, + "qwen_elapsed_seconds": 5.184, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_0d4dcd9caae2/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..bf212e0cb2b1177e373500a9d032be809d5fe86a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4722c68ee255fba4f7fdb2ebd5627e6c41b0e89306a4956b94e83e370c5f7edb +size 469231 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..77dbe42f9dce63ee64f1057c5d16c4c7bca3e0c4 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f5c882ab34fbf4b7704e2f30d053a193c576f62ada04c97abec54883cf38ede3 +size 602521 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..385fac9a3fa4ded52288c60c0796584410111b3c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:385e7547df92da7af1aafba894dacfa8cff6f7ce750769d8637642ea7315a1e8 +size 519378 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/record.json new file mode 100644 index 0000000000000000000000000000000000000000..0a1650beea34482f9b1c0c46c761a08e7ad64a77 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/record.json @@ -0,0 +1,66 @@ +{ + "prompt_id": "multi_object_relations_d4_11d2a4093355", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic image of a box with a book inside, a mug on the table, and a sign next to the box, high contrast, clear foreground separation", + "target_evidence": "book inside box", + "expected_visual_elements": [ + "box", + "book", + "mug", + "sign", + "table" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", + "diversity_tags": [ + "inside", + "box", + "table" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "11d2a4093355", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.5769, + "diversity_score": 0.2231, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_11d2a4093355", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_11d2a4093355/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23095593, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6489438, + "qwen_elapsed_seconds": 5.117, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_11d2a4093355/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..de0d88f7a7ecacd29173cae3f272f8446004b00b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad45410654a2bd20565047df6baff1aeaf2462f0bec1e9a6da3c6831d6bf520f +size 242431 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..db9eb6a6b00ce071dc8461515228e5c7a4bc204c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed50a2027bb20d311a8704f5ea69da0fdd02c29289be7f56e76424dd2376c1e5 +size 474243 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..0c8a40884cf3f4e2f6a55e5b32028bf3255bb2ae --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4dac0ed3f3de97fd9584dce8582ed3340c72f256c5bc0f5b17490199cd50ed67 +size 518508 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/record.json new file mode 100644 index 0000000000000000000000000000000000000000..1626f9fdb3aecb3a8d4c90101e53f1527c301f32 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d4_1b6fe1b04f06", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic setup with a tool placed on top of a box on a shelf, clear foreground separation and high contrast", + "target_evidence": "tool on top of box", + "expected_visual_elements": [ + "tool", + "box", + "shelf" + ], + "negative_prompt": "distorted perspective, low contrast, unclear edges", + "diversity_tags": [ + "on top of", + "box", + "shelf" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "1b6fe1b04f06", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.8, + "diversity_score": 0.0, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_1b6fe1b04f06", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29886781, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5979193, + "qwen_elapsed_seconds": 5.188, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_1b6fe1b04f06/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..3deea21a27420f0e68b2dc83301bcf367e29472b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5752aea297d21d6c509493a6b34103112aaf03b319a30ad07e72bcff3101cfb6 +size 555273 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..e472df5cb64c9f06d18134566436a493b75fbe37 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e2e224bc655c864a5cb7167b7ca64afc3cfb90a5227458d7127ba08754d77220 +size 546768 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..303743fc92b8e0450283dba63ec7898d3abf7429 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:852e8929ca3274d59db43be56985e9ff7760b0839904cc4e15dcd5751aee7334 +size 568013 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/record.json new file mode 100644 index 0000000000000000000000000000000000000000..bd9b42ade5ec5a7860d46dfed56dd779e99ced63 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d4_244ce55796c6", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic setup with a box on top of a shelf, high contrast, clean boundaries, and distinct foreground objects.", + "target_evidence": "box on top of shelf", + "expected_visual_elements": [ + "box", + "shelf" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, real people", + "diversity_tags": [ + "on top of", + "shelf" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "244ce55796c6", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.6364, + "diversity_score": 0.1636, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_244ce55796c6", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_244ce55796c6/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29678577, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2689398, + "qwen_elapsed_seconds": 5.178, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_244ce55796c6/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..9c0c502609f777153dce267ba11cc18f7a4c22fc --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a8818f456f7d990d7251424c72cb2109fe4fb62ed9f06aaa3bb374dce64e00e +size 363002 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..07bac601e4dee1e602dea18526bf4b81aef5243f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b3e8450e6e58426b47db6e57d8970d601307fce376b9a7bea05b2ccc9b9cfe85 +size 659721 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..155d224922b869ecf848136da716b5074b1a6b74 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f1aa02e02140b562a69fe828393e65749b3dac2d84b2095dcb4177966cf12bd9 +size 580511 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/record.json new file mode 100644 index 0000000000000000000000000000000000000000..ebe0b49b40eae0a14e1dc0841e525f5fbe3284b0 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d4_3bf4f5d535e5", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic image of a mannequin with a bag on its shoulder, a pen inside the bag, and a sign near the mannequin, clean boundaries, sharp focus", + "target_evidence": "pen inside bag", + "expected_visual_elements": [ + "mannequin", + "bag", + "pen", + "sign" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", + "diversity_tags": [ + "inside", + "bag", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "3bf4f5d535e5", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.6364, + "diversity_score": 0.1636, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_3bf4f5d535e5", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29156616, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7859375, + "qwen_elapsed_seconds": 5.232, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_3bf4f5d535e5/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..46df0c220c90d2dc76c28b731b6da7b858446336 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb3317451697fe71ee411b6f6604e1d5a8a3075e007246d83985f0ae4bb695f9 +size 529679 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..74a059b814c8a4eae52eafcf608a400f291e8ed4 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44ebb1dd743061885326acde546108493bbf765a991dc6b33cca444b29148901 +size 537150 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..425fa72e05950e4638caf53add1dcf6044f9b589 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:29031ad8ac9c2816d402cda0e7dbe5ad139ff2d0381cd425fcecd7226ff397e9 +size 534386 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/record.json new file mode 100644 index 0000000000000000000000000000000000000000..d09a2915dcd860d77e938fdd59b9973b545f0772 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d4_3d1b4c30ae72", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic scene with a mannequin holding a toolbox, a wrench clearly in its grip, and a screwdriver nearby on a shelf, clean boundaries, sharp focus", + "target_evidence": "mannequin holding toolbox", + "expected_visual_elements": [ + "mannequin", + "toolbox", + "wrench", + "shelf" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", + "diversity_tags": [ + "holding", + "mannequin", + "shelf" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "3d1b4c30ae72", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.56, + "diversity_score": 0.24, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_3d1b4c30ae72", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 22082823, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6994313, + "qwen_elapsed_seconds": 5.143, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_3d1b4c30ae72/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..32f1a35e49a6e7298b4c59e9a24dac89af61ad1e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:54bc0bd5530efad58d0572fc5cd4575b1e5a15c261103bb3965b7e3e53ee59f5 +size 394464 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..afba07256b8d438d3cd94ec09e3663d08e1b2124 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b5584f1b0b8f19cd73c3a13441476d9112eace3db2f0939a22e1c09e366a91f6 +size 575678 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..b0722e4ef9c80a6a066b755581ee91b619249ddb --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:595ff7e7e58cb13663f126c04d9c7c4ce1f53d9a2243672ce2ea7fb449cca855 +size 575383 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/record.json new file mode 100644 index 0000000000000000000000000000000000000000..339a7d5c5ccb329cefdbc2f766d3cb333448ee56 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d4_541f29108785", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic setup with a mannequin standing behind a chair, clean edges and sharp focus to emphasize the behind relation", + "target_evidence": "mannequin behind chair", + "expected_visual_elements": [ + "mannequin", + "chair", + "floor" + ], + "negative_prompt": "overlapping elements, poor lighting, low contrast", + "diversity_tags": [ + "behind", + "chair", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "541f29108785", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.48, + "diversity_score": 0.32, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_541f29108785", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_541f29108785/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 24655905, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2320017, + "qwen_elapsed_seconds": 5.142, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_541f29108785/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..3393401a69fab5d2abafb3d5a5c1f597800ae64f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7c3b501e2deec6273f703b1df72c30e1960e30d583fa226b572b2d71176b8776 +size 413050 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..8ba8033fd5e992e7014bc4c5d7ddb07597a9d9cd --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c8b6ba8ecf08891aec5a7a6153022afb3cb0d4b589756949f08ada7fafb809b1 +size 650636 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..11ea9fa41f2c3ea2d33dfe2833cf29c63e744a02 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa32d7ffd4302641ad1f8609b4f596818a8d0efe1e29962ae4a8103755f99274 +size 604965 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/record.json new file mode 100644 index 0000000000000000000000000000000000000000..3c71fa906b40de27c2032b2938b1c2ef0f94aaae --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/record.json @@ -0,0 +1,66 @@ +{ + "prompt_id": "multi_object_relations_d4_57a5e1f493b5", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic scene with a mannequin holding a bag, a key attached to the bag, and a tool on the desk, clean boundaries, sharp focus", + "target_evidence": "mannequin holding bag", + "expected_visual_elements": [ + "mannequin", + "bag", + "key", + "tool", + "desk" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", + "diversity_tags": [ + "holding", + "mannequin", + "desk" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "57a5e1f493b5", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.7143, + "diversity_score": 0.0857, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_57a5e1f493b5", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 27538672, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 3645504, + "qwen_elapsed_seconds": 5.219, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_57a5e1f493b5/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..6787ad7a662eefda93a6113aa3638c369c25c2f0 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:51c10c5eaf4228631e59229f0c6419336486f0261d48585ddcd218492cea8b4a +size 390366 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..05214aa0a8b0e1756e27ce48173bbf6ae50a2041 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:05f098be77fb0136a86427b8f8d75b97db2ecf8659cbd92b9ad470451a094dfe +size 562115 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..19bad3ba8e97211915c47f2e0deae5fc2d1bc475 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e54fc3d7bb98a5318674ff01c53e2909e2636aa01c5b06b149da38fd692f9679 +size 589893 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/record.json new file mode 100644 index 0000000000000000000000000000000000000000..fe872abd7e36118c0a392b513ab48231a4797f55 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d4_58ff6e8b16b1", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic view of a mannequin with a tool attached to its arm, clean boundaries, high contrast, and clear foreground.", + "target_evidence": "tool attached to mannequin", + "expected_visual_elements": [ + "mannequin", + "tool" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, real people", + "diversity_tags": [ + "attached to", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "58ff6e8b16b1", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic view of a", + "max_similarity_to_other_prompt": 0.75, + "diversity_score": 0.05, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_58ff6e8b16b1", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26440915, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9559365, + "qwen_elapsed_seconds": 5.213, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_58ff6e8b16b1/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..7afee88e71978773e5b89dded76ae523e43a142c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e38b601b701581d2868fa55eaf75fe0200502352b45b03e4a5588be9bfeb2ee +size 469650 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..3a59f4128e5e0cfb2b82c0e41f5256b208a3d4b8 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b06ff1f27c71c2c3cb915532908bceb7a381641fbf3c271917887b473cbd8fd4 +size 491411 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..b5806a1e5fe89c1734d688482664fae6b515a2ec --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:267f14059f19f0540f292097b7f6e548b0c2416c754d5fe98d0ddcb0c4604897 +size 544397 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/record.json new file mode 100644 index 0000000000000000000000000000000000000000..6da6ae25165da53d6e4f8834e611ade2d066d6d0 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/record.json @@ -0,0 +1,62 @@ +{ + "prompt_id": "multi_object_relations_d4_5b222069f69d", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic view of a box on a desk with a tool nearby, distinct foreground separation, and high contrast.", + "target_evidence": "box on desk", + "expected_visual_elements": [ + "box", + "desk" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, real people", + "diversity_tags": [ + "on top of", + "desk" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "5b222069f69d", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic view of a", + "max_similarity_to_other_prompt": 0.619, + "diversity_score": 0.181, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_5b222069f69d", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_5b222069f69d/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 22137842, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9803052, + "qwen_elapsed_seconds": 5.142, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_5b222069f69d/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..8679485a803e00d8ff451c2b2ca560844fcd3723 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a9f5e6de7f456506c9e22649575ac23ab9b67595129cfaa705ebdb778e45f82a +size 63268 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..7bfb07e2342497d8c1ffadec9ca68a34f2cd5ce1 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67b49c78e95b3ee110b131ca1de58ee21b4cea88b1ff2f6a7b4682818395248c +size 443390 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..4e804075bab7d1997bbd8025bc4f9c05c01ba63f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:30b2f89cf992c7d74b551a9f4ed371df6672657931bb3a1c0cc9f80ad8e671cd +size 590574 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/record.json new file mode 100644 index 0000000000000000000000000000000000000000..214be4118279ec1a5407e279fc965c246ae1a346 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d4_62ec93cecea4", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic setup with a tool on a shelf, a wrench on top of the tool, and a bag near the shelf, clean boundaries, sharp focus", + "target_evidence": "wrench on top of tool", + "expected_visual_elements": [ + "tool", + "wrench", + "bag", + "shelf" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", + "diversity_tags": [ + "on top of", + "tool", + "shelf" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "62ec93cecea4", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.619, + "diversity_score": 0.181, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_62ec93cecea4", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_62ec93cecea4/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20476669, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6514423, + "qwen_elapsed_seconds": 5.125, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_62ec93cecea4/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..9eb309f5329950dd11fc5eb494eae706a2e149c2 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd9423b4eea25e6c2ad98a54958ae06e8c5e3d8f6541c7d82cd7fed6d10c2a2d +size 418308 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..4ee1b284dbac7e76e84287f1e8155d943546308f --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a21391dd087c2e23e58e1ade8a11370fde52b5ec592dd77e1027c77ddf3ca802 +size 540170 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..0261617936eecdca1687439e91967e6efe605a65 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a25b440af6bc29d08561d0acf7f3bfbaf356e3c7efa18ff6b58cff4ffdf4dc0 +size 598932 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/record.json new file mode 100644 index 0000000000000000000000000000000000000000..6afab3746785a471060d20d77c0c2f14c1e00b8c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d4_64ae9cc06dca", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic setup with a bag placed inside a container on a desk, high contrast and clear foreground separation for relation analysis", + "target_evidence": "bag inside container", + "expected_visual_elements": [ + "bag", + "container", + "desk" + ], + "negative_prompt": "distorted perspective, unclear boundaries, reflections", + "diversity_tags": [ + "inside", + "container", + "desk" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "64ae9cc06dca", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.7273, + "diversity_score": 0.0727, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_64ae9cc06dca", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26751141, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5779002, + "qwen_elapsed_seconds": 5.15, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_64ae9cc06dca/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..c33433afeb555a1931693440e944d6fe4441b29b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ecba5b4d45c8edd4578b9993728cc659a07a1743422a9f2ade118f3f977f6ef +size 458394 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..a610e46ef0c12a28250fa79638d4aa542a53e73c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:54745e2d009d61fe2517cdc93c8c8705d645d1774c0426fd10e766d585b86154 +size 583125 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..975b2f85d8104a2d9a01fbf4a381468aa05bdfb7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd57acd9f791d8acfa5ac3433b17fb3e42a592f33ecf2e49e79052236d4335cb +size 579180 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/record.json new file mode 100644 index 0000000000000000000000000000000000000000..68d189e383ca9c53bb0caf0f21211de05a127afc --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d4_6ad705a61b48", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic scene with a book on top of a box on a table, clear foreground separation and high contrast for relation inspection", + "target_evidence": "book on top of box", + "expected_visual_elements": [ + "book", + "box", + "table" + ], + "negative_prompt": "faded colors, background clutter, unclear lighting", + "diversity_tags": [ + "on top of", + "box", + "table" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "6ad705a61b48", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.625, + "diversity_score": 0.175, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_6ad705a61b48", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_6ad705a61b48/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26682356, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5793693, + "qwen_elapsed_seconds": 5.175, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_6ad705a61b48/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..4eec38944a2b48ea57515b4b7b6495511fcf87cc --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9bb1f16fea220594511c1fd710c4a0ae6b863b4d7b486087590dfee181950cf +size 501874 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..681ce3f166dfafbe922f9bfe349f319b1883493b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bcf0d363c6d21a0a80d55ff81abe99b2dc6ae4cb872ace626bde2f4bdea90fd8 +size 554901 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..652adf6239b1626cf148f9ea9b49d2e0a71322a1 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e74434cef5666243ecab43f60c156e983b8d2addd236d3df7b2b93a81258491a +size 543323 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/record.json new file mode 100644 index 0000000000000000000000000000000000000000..5fe6903149ca6cfaa95b748f53ab1c05e19efa7a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/record.json @@ -0,0 +1,66 @@ +{ + "prompt_id": "multi_object_relations_d4_6bdcb20a1fd4", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic scene with a mannequin standing behind a chair, a lamp on the desk, and a cup on the chair, clear foreground separation, high contrast", + "target_evidence": "mannequin behind chair", + "expected_visual_elements": [ + "mannequin", + "chair", + "lamp", + "cup", + "desk" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", + "diversity_tags": [ + "behind", + "mannequin", + "chair" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "6bdcb20a1fd4", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.6087, + "diversity_score": 0.1913, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_6bdcb20a1fd4", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23139165, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2585804, + "qwen_elapsed_seconds": 5.112, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_6bdcb20a1fd4/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..3c8a7523b62708c5d3dbe75cbc17f503a3d04a41 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:04346a8d6f66fc87685c9dd7946e04cbdd4727c940a50fb3846d2827fe7af403 +size 458952 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..2faf41e77c58e0fe77449df5b0933a6e7cebda00 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2577e5f67abc9d8528500c383aa6e1205684f6b02c1201a1ba063dac846cd440 +size 539838 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..d10ec9b0fe23dc10127aa49e927ac782d2f50467 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c22875f5f9865ddf16ffab57dcbf8be8e6ffd2eb4b8b172fa07bbd6dcf3e29e +size 538025 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/record.json new file mode 100644 index 0000000000000000000000000000000000000000..03a0b9ad4c9bba8199b5fd94c62c2fd62bd40d14 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d4_8ceda8b84aa3", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic setup with a container on a shelf, a bottle inside the container, and a tool on the shelf, high contrast, clear foreground separation", + "target_evidence": "bottle inside container", + "expected_visual_elements": [ + "container", + "bottle", + "tool", + "shelf" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", + "diversity_tags": [ + "inside", + "container", + "shelf" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "8ceda8b84aa3", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.6364, + "diversity_score": 0.1636, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_8ceda8b84aa3", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28209408, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9351229, + "qwen_elapsed_seconds": 5.178, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_8ceda8b84aa3/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..352386792e3f10f8cb213347243c454d94451f1a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1842791b32f0feeded453f202fae0307996047f318189f745204cf62be159aed +size 404196 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..864eb2a9f25eac3b295ce6bac4f930265dfcad3d --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67481246149fca422ccdadaa1515181e89b51d8a6d079635c128ab908a9476a4 +size 361223 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..87a0220be7ef458320bde603664da35aa48b85c7 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3857b127a5ca47807ab79f69a117c35f1952e3a9a5cb562beed7393ab07bd92e +size 641974 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/record.json new file mode 100644 index 0000000000000000000000000000000000000000..68cd6fdc2dd15a6b86e3fb5579a038018e11a53d --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d4_a48ec906a073", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic scene with a bag on a mannequin's shoulder, clean separation and sharp focus for the held-by relation", + "target_evidence": "bag on mannequin's shoulder", + "expected_visual_elements": [ + "bag", + "mannequin", + "shoulder" + ], + "negative_prompt": "blurred edges, low resolution, cluttered background", + "diversity_tags": [ + "held-by", + "mannequin", + "bag" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "a48ec906a073", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.6, + "diversity_score": 0.2, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_a48ec906a073", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_a48ec906a073/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29554811, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5188663, + "qwen_elapsed_seconds": 5.098, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_a48ec906a073/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..1530d4507aac011b1c34d2cc7c285119a1202909 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95a261a35f8a80bd32f5cc10a00ecbcecc58f22b422fc23028363d6bc2f02702 +size 470522 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..8655d881323158d6b8b2b917b4ab7cfbe79a992b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b35968f29b732d5536ce4d7aaf16b4dcecfddec7fe7ea72b32e4a90f71bdd8e7 +size 496614 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..04490038eed264723c69ec089b1a4d269407d527 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca6ca034cd42f9a39810f57c148a0e8c29a31d8f5f4d275079d5a35fd403149e +size 528924 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/record.json new file mode 100644 index 0000000000000000000000000000000000000000..0dd16163e276316dae8f29f4f17b31eacbaea7f1 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d4_dc8bf3cb3eab", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic scene with a mannequin holding a box on a shelf, clean boundaries and sharp focus for clear relation inspection", + "target_evidence": "mannequin holding box", + "expected_visual_elements": [ + "mannequin", + "box", + "shelf" + ], + "negative_prompt": "blurry details, overlapping objects, low contrast", + "diversity_tags": [ + "holding", + "shelf", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "dc8bf3cb3eab", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.7273, + "diversity_score": 0.0727, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_dc8bf3cb3eab", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 22020640, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 7567140, + "qwen_elapsed_seconds": 5.225, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_dc8bf3cb3eab/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..94bd143432298b77790d1e9e54466982ed7ce5ec --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3668dea18c7afa52937058485141b70d187305079a31227ddef60c4b7174c747 +size 530787 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..8bc13cb988dc01b9366913190dfc602c4cf2bf48 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:416d9c3dc1816c61cab59cbc6d5d52520d3408cbe021fc770b21ea3554800f98 +size 565060 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..c67e1ba5b997cc4e4f5e590f81eef35312f6595e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1718f94ea1227cef5106389eccf63b7610a2c20467358f447fb26fa5336d2cac +size 572842 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/record.json new file mode 100644 index 0000000000000000000000000000000000000000..827b8875b69599ed48ad0b53244b0a1ed6259701 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d4_e00ad2c64904", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic scene with a container inside a bag on a desk, clear boundaries and high contrast for relation analysis", + "target_evidence": "container inside bag", + "expected_visual_elements": [ + "container", + "bag", + "desk" + ], + "negative_prompt": "distorted perspective, low contrast, unclear edges", + "diversity_tags": [ + "inside", + "bag", + "desk" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "e00ad2c64904", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.7273, + "diversity_score": 0.0727, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_e00ad2c64904", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_e00ad2c64904/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28208223, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 9235420, + "qwen_elapsed_seconds": 5.099, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_e00ad2c64904/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..327867adf73ab44064f2d0899d664ac99abf8ff9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3fe4bdc20e169e67ab7f92653349de1b6fcf42e4953bddebb69c05837fa4dd21 +size 436801 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..8d48fa704fc5ed0b7738ef5fe6804b1b290a79cf --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d91ca0eb2812acaa52e931363b7d41e016fa4f1001b1ffa7a8f70d44de79881 +size 567755 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..3a6d5c236388f9a67e2845169882c3ddaece77fd --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1ec7c9294f93799eeecce7690b1ba3d8bccbb69be5699513882d449b86c341e +size 580332 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/record.json new file mode 100644 index 0000000000000000000000000000000000000000..51dd84ce3fc064a2867308c685e460cd1b1f52bd --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/record.json @@ -0,0 +1,66 @@ +{ + "prompt_id": "multi_object_relations_d4_e2ae8a1373cb", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 4, + "image_prompt": "synthetic photorealistic image of a box with a key attached to its side, a small plant on the table, and a sign next to the box, clean boundaries, sharp focus", + "target_evidence": "key attached to box", + "expected_visual_elements": [ + "box", + "key", + "plant", + "sign", + "table" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, unrealistic lighting", + "diversity_tags": [ + "attached-to", + "box", + "table" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "e2ae8a1373cb", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.5, + "diversity_score": 0.3, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d4_e2ae8a1373cb", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26753698, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 161678, + "qwen_elapsed_seconds": 5.158, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d4_e2ae8a1373cb/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..e790ecd8fefac8a9ff80ae4974eefdcb5f0ecfe6 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be4124077000a1432b500320127d6c8a6f8afa6afca2e6369d8f292a2f36eaae +size 457514 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..1b2522ce8f020399998bf8b0eefa0bbacdb73bef --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:97366238a1340498485cd619da9eaaf484950c2600fe538919bd2997a4e5bda2 +size 568443 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..7cb71e56f53a864e40592ab51214eee40d9cd3cf --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e3cdd0f4f9ed26ec765a96b97c54d93932422a70f6e6cb0c4b3de1cc7fde9b7 +size 585455 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/record.json new file mode 100644 index 0000000000000000000000000000000000000000..e3ded020ad7df7c055da5b6adb1ccfa3e829e8cc --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d5_11fc8e7a0239", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "synthetic photorealistic image of a bag under a chair, with a bookshelf in the background, clear foreground separation", + "target_evidence": "bag under chair", + "expected_visual_elements": [ + "bag", + "chair", + "bookshelf", + "clear foreground" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "under", + "bag", + "furniture" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "11fc8e7a0239", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic image of a", + "max_similarity_to_other_prompt": 0.4091, + "diversity_score": 0.3909, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d5_11fc8e7a0239", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25246698, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2451426, + "qwen_elapsed_seconds": 5.167, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_11fc8e7a0239/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..48bbae84300f5eb97fae06ac7bac6fae1fe4928e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:43e53bfb0ccca41952cadf93e34056f85b46b2909ebc7030a71d9694b1d2e1f6 +size 289084 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..701a350a6d7e06c43792454b1a8b97726ee4b81c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85d0a77bbaa7d173dd7e06b8a03dd35be7a67570ad4b63ce465d17c308a9dfa8 +size 555997 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..8caba1cef19c5686ad51816bde2266c00e11d768 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:12d0c1582f6755096d483daf51de7c9efb94c10cd43a19357b61359524bf98b9 +size 583839 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/record.json new file mode 100644 index 0000000000000000000000000000000000000000..234fd81c18ecf14a1b53382ee793ead9bce2fd79 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d5_34c2ccf3b5ef", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic setup with a bag placed on top of a table, clean boundaries and high contrast for visual clarity.", + "target_evidence": "bag on table", + "expected_visual_elements": [ + "bag", + "table", + "clear boundaries" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, real people, faces", + "diversity_tags": [ + "on top of", + "table", + "bag" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "34c2ccf3b5ef", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.5909, + "diversity_score": 0.2091, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d5_34c2ccf3b5ef", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 22483845, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2615726, + "qwen_elapsed_seconds": 5.109, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_34c2ccf3b5ef/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..b2decb25eebba93dc3bbc4cc4aa36189abfbb904 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5975d91d13e8ce2b028c36746d9cad5fb5fa3ad0d2e066c481ec16f4f98c9c9c +size 457719 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..5b5b44b8a396e4aa3f2544ea17ef79ccadc31ab5 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b8e2cda445368c9e8ce3e23a7e22783d7872eca462cf51a8cde8cdc8c6404049 +size 556289 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..48120303f8720db409e8e2ccfb01f74b281b22e2 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0214d38e8c74b47c3e18379d9b62eff303998780d0fb9cd0ef6f6b3b7bdadfa0 +size 569772 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/record.json new file mode 100644 index 0000000000000000000000000000000000000000..f76d8ac92c325b6a3c6768d35d44090b2f0ed129 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d5_4f451cd1db1d", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "synthetic photorealistic setup with a box on top of a shelf, a sign next to it, sharp focus on the relation between them", + "target_evidence": "box on top of shelf", + "expected_visual_elements": [ + "box", + "shelf", + "sign", + "sharp focus" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "on top of", + "box", + "shelf" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "4f451cd1db1d", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.6667, + "diversity_score": 0.1333, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d5_4f451cd1db1d", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 26509252, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2735174, + "qwen_elapsed_seconds": 5.158, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_4f451cd1db1d/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..e5195ccfaf31efdd61faa1b2079abdeec04ab12e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:94f1e9957709498652884b7b5af6dfee6b9f99744b0785f1cc814f03377685bc +size 515723 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..9f153b63e5dbb13ea1d35295d9097930ef973d67 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:92b19ef7e0567326a65f0d35432c2e87fa7ef3320a64ecfcbb1a578bec686bdb +size 528666 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..acbc8486103d0f99a7cb106392077f3ac51d6b3b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5fc872210773c9beeb46e16cb61132dd7ae9970fbca085ab86eb25140ea03032 +size 602110 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/record.json new file mode 100644 index 0000000000000000000000000000000000000000..22575aeb58ce410ac2fc8b700897bd5b3636b797 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d5_5db805d7dab7", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "synthetic photorealistic setup with a tool on top of a mannequin's head, a box beside it, high contrast for visual clarity", + "target_evidence": "tool on top of mannequin", + "expected_visual_elements": [ + "tool", + "mannequin", + "box", + "high contrast" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "on top of", + "tool", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "5db805d7dab7", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic setup with a", + "max_similarity_to_other_prompt": 0.5217, + "diversity_score": 0.2783, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d5_5db805d7dab7", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_5db805d7dab7/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20942630, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5378640, + "qwen_elapsed_seconds": 5.126, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_5db805d7dab7/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..26c41ed3ec0eaa33180f96293ae17c6d837b8567 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1029478603f905d2f738c128037e742373bfaa8c704cde0fb635a04baede266c +size 469662 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..69f5dd3c9e5b39542945fdb15e83c675b23513ac --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5121883d696129f42f64467d13c740bd10351b50fba884cf1ab09d27a9fc1da8 +size 594730 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..49fb35ffc5778908e227ce411ee32f206e24ca1a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a5fb1496291e7ef15927ff3eccded984a6438825a3b4fd335340ce89421a62b5 +size 603377 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/record.json new file mode 100644 index 0000000000000000000000000000000000000000..7adb304e7dda6c9fff06851f3c7fa389e3713d20 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d5_5eae587b509d", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic view with a mannequin with a bag attached to its arm, clean boundaries and high contrast.", + "target_evidence": "bag attached to mannequin", + "expected_visual_elements": [ + "mannequin", + "bag", + "clean boundaries" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, real people, faces", + "diversity_tags": [ + "attached to", + "mannequin", + "bag" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "5eae587b509d", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic view with a", + "max_similarity_to_other_prompt": 0.75, + "diversity_score": 0.25, + "repeated_opening": false, + "review": { + "prompt_id": "multi_object_relations_d5_5eae587b509d", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_5eae587b509d/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 21052723, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 2471002, + "qwen_elapsed_seconds": 5.111, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eae587b509d/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..7b0c84d2c96865239c5779e806fddb96131a75c6 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:386a97270a4086ca764b9badc9027bcdb7757e7bb545e1b995cfff4399ef0b3c +size 457920 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..bb5882057b296e099898288cadeb513c770b59de --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:821587aab2f0231b7ff6e46f4deb4b06a3c0349e36f29dcad9cd509943c5635f +size 557886 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..e49ee5f10eef130abba063d497ada18ef507a47d --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4752d6f2370ed10605189815e693d17caf43565ebcd96d3ed44c12e48d041610 +size 587616 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/record.json new file mode 100644 index 0000000000000000000000000000000000000000..b589daa809707add87247b080ff4fae5548c0221 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d5_5eeca9085072", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic scene with a mannequin holding a box on a shelf, clear foreground separation and sharp focus.", + "target_evidence": "mannequin holding box", + "expected_visual_elements": [ + "mannequin", + "box", + "shelf" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, real people, faces", + "diversity_tags": [ + "holding", + "shelf", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "5eeca9085072", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.7368, + "diversity_score": 0.0632, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d5_5eeca9085072", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_5eeca9085072/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 27573616, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 944517, + "qwen_elapsed_seconds": 5.115, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_5eeca9085072/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..fec6d7122b1d95676378e3a0e5297132ec3dd26c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb3d1b68607f53cb0c36338c43041801a92763c7e278618bbb446b53a7ebb590 +size 529178 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..3838d16d3d3ae3093ef491e15ee9bf8ee57c2057 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e073c7af0b127f2dae6569729fbff7a86c94d0aca1719dfca003056e75d0603 +size 592884 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..5107bfe8e174fcee5f42c2a660595bfcfb0897f1 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf79cb95ad1463e2c9a8904cf50ed1b01d6fbe41d50efde46a6d9c43d70126c4 +size 587807 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/record.json new file mode 100644 index 0000000000000000000000000000000000000000..26a6cb3b2ef66eb036831ddf1ba0f15a4780e41e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d5_63e8915777db", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "synthetic photorealistic scene with a mannequin standing next to a box with clear visual legibility and sharp focus.", + "target_evidence": "mannequin next to box", + "expected_visual_elements": [ + "mannequin", + "box", + "sharp focus" + ], + "negative_prompt": "low contrast, ambiguous placement, soft edges", + "diversity_tags": [ + "next to", + "box", + "sharp focus" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "63e8915777db", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.5238, + "diversity_score": 0.2762, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d5_63e8915777db", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_63e8915777db/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 28877290, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 5370677, + "qwen_elapsed_seconds": 5.181, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_63e8915777db/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..0f08a7d9eb4ddbae64cd0f5b8de44027eff4c998 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a2a02d098d069752eb5f3bb2530c46d18ef5e5454fbba6a4d93422342e6d344 +size 373216 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..ad2886ffd3d1acd2fdcb6e952a53c379c7205d3e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6dc5d3cb40284ef40c7f8138ae81c382ae27cac64e78ea44a2e29a2ec8516f9a +size 512655 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..8e89e4f14dabfc9c4f7ecf9d70338620ffde538e --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e7e0d58f27b5144a846944b6476cd6fd97115a80998132c471bd5c645ed193a2 +size 549164 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/record.json new file mode 100644 index 0000000000000000000000000000000000000000..ecbe86c3587c4f68d07deb5f8f17742ec654ae74 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d5_6e43b96c2267", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "synthetic photorealistic scene with a container behind a desk, a tool on the desk, clear foreground separation for relation", + "target_evidence": "container behind desk", + "expected_visual_elements": [ + "container", + "desk", + "tool", + "clear separation" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "behind", + "container", + "desk" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "6e43b96c2267", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.5, + "diversity_score": 0.3, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d5_6e43b96c2267", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_6e43b96c2267/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29377294, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 735298, + "qwen_elapsed_seconds": 5.157, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_6e43b96c2267/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..7f6597864f495ebe7dfd7f4020bdbdf3a66e302a --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c619cae89658744507eb6be72312038833db22c43ced03cdae5a0f121d5104b9 +size 454736 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..9fad0cfcf052eaee6d35f5cdd4c0fc26b148155c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a530e803959f9d722abec5dd0c3720dfccf3877d5ca052352173b6aca5717516 +size 581182 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..c41adefe552c2339881e155390e2280d481a99f9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:90efd1a8b305890f928273ac75cf288ee602b4457a0e5bccfe546730f5961da2 +size 616452 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/record.json new file mode 100644 index 0000000000000000000000000000000000000000..22ed125d50c0b9e145467d5b7fd2ed56bb64e1cf --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/record.json @@ -0,0 +1,66 @@ +{ + "prompt_id": "multi_object_relations_d5_97c8654f9f45", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic view of a mannequin holding a toolbox, with a wrench on top of the toolbox and a screwdriver inside the toolbox, clean boundaries and sharp focus.", + "target_evidence": "mannequin holding toolbox with wrench on top and screwdriver inside", + "expected_visual_elements": [ + "mannequin", + "toolbox", + "wrench", + "screwdriver" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks, real people, faces", + "diversity_tags": [ + "holding", + "on top of", + "inside", + "mannequin" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "97c8654f9f45", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic view of a", + "max_similarity_to_other_prompt": 0.6087, + "diversity_score": 0.1913, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d5_97c8654f9f45", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_97c8654f9f45/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 20498959, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 4556878, + "qwen_elapsed_seconds": 5.194, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_97c8654f9f45/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..1683c9f15cc5bcac8017c29ff832752ec1439b87 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f95f527bc75379bd6c0e6b728a122a4b7cd9d2c13d13c69b12df121a658d077 +size 318306 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..04b70ac27bff068e4ff6094470fe11af90c17050 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ad502273f5c1bbacb2adac7762c5cd8b1313c659a0003e171ee967fc9ded916 +size 594164 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..7f87658be9182604725264616bde13aec1f7c359 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c342e77b0dff1d7d1e401d4deb9fdfbfd5cee4ca68abebb1e8041c3c93c6ea18 +size 616591 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/record.json new file mode 100644 index 0000000000000000000000000000000000000000..1930e5d21d32fc9f0c73fc056eda4a96bb6f5943 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d5_98689667028e", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "synthetic photorealistic view of a tool resting on top of a box with clear visual separation and high contrast.", + "target_evidence": "tool on top of box", + "expected_visual_elements": [ + "tool", + "box", + "high contrast" + ], + "negative_prompt": "low detail, overlapping elements, soft focus", + "diversity_tags": [ + "on top of", + "box", + "high contrast" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "98689667028e", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic view of a", + "max_similarity_to_other_prompt": 0.6667, + "diversity_score": 0.1333, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d5_98689667028e", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_98689667028e/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 25981097, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6143210, + "qwen_elapsed_seconds": 5.118, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_98689667028e/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..cc4b1e48de0b110cfa7746defb5c63dfb3a4b3d1 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:28e0fe465999f8c8da893bef0d4d0d7837de7e4ba62f68745be593f3dc815889 +size 749442 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..617422b94cdc03e8ed2b12331bf4889b523d29a6 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd95b788acbf0ca66be1987fc54629e7d82d1ee5c9809683104cc3901e33cc0b +size 560961 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..63631921a07f84409eed8d62e462d7ed482b4143 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f5cb0a4d2fd342e66b41749383a86cdbdfa0e5cb1fd7e921fec532caf992b57 +size 559006 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/record.json new file mode 100644 index 0000000000000000000000000000000000000000..e0beaeebd0db687bc21b15ee29e2965ec9265d4b --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/record.json @@ -0,0 +1,64 @@ +{ + "prompt_id": "multi_object_relations_d5_99ab20067c80", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic scene with a sign attached to a wall, high contrast and clean edges for visual separation.", + "target_evidence": "sign attached to wall", + "expected_visual_elements": [ + "sign", + "wall", + "clean edges" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, real people, faces", + "diversity_tags": [ + "attached to", + "wall", + "sign" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "99ab20067c80", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.75, + "diversity_score": 0.05, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d5_99ab20067c80", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_99ab20067c80/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 23192110, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 6069800, + "qwen_elapsed_seconds": 5.205, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_99ab20067c80/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..63e134b3003cd6972274b7710c5c73a0052a159c --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:26b1de032c211825f99c719819f5186e3807ba0f16a033e7416d5d530d6224bd +size 375504 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..4f191b97ea50a7226131286f46c20b97077950fe --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a84ec6fd82883d55a6534074ff73078de4a48393c103e52193c5819bc16a879 +size 557597 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..2fa71d8bb737bd5f66bd193fb658be5fb2720b64 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3711a0624be72f82c9191ec46efe61526881b4ed13dda3ac0db369910a2d59ed +size 599181 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/record.json new file mode 100644 index 0000000000000000000000000000000000000000..8bf6b342c6825c7f618128d2761395e5c3adbacc --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/record.json @@ -0,0 +1,65 @@ +{ + "prompt_id": "multi_object_relations_d5_a2bed91d138b", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "synthetic photorealistic view of a mannequin with a sign attached to its arm, a tool on the ground nearby, high contrast for clarity", + "target_evidence": "sign attached to mannequin", + "expected_visual_elements": [ + "mannequin", + "sign", + "tool", + "high contrast" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks", + "diversity_tags": [ + "attached to", + "mannequin", + "tool" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "a2bed91d138b", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic view of a", + "max_similarity_to_other_prompt": 0.64, + "diversity_score": 0.16, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d5_a2bed91d138b", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 4, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 0.97, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_a2bed91d138b/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 29882024, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 298442, + "qwen_elapsed_seconds": 5.139, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_a2bed91d138b/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..1423d95d7555b7f3db3a36f392616f854bcfa186 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dcea695da087c45a08f615dce03d9dc8951cfa88a3a3533512a3489c8c89601e +size 428743 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..3ac54aa624a4badf1a8ed631f3374bfbc0378b94 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:79616a52e3647e9f6bc313ec1c283ce9c61c8630cbd112f16c156be19183e681 +size 544010 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_02.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_02.png new file mode 100644 index 0000000000000000000000000000000000000000..165917dc5903ab8bff6305ed0abbf03569272575 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_02.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:49e3d0b2d688f1366e837bc9a1547152ce36a0e118bc674d7fa2dbc698d00861 +size 656356 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/record.json b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/record.json new file mode 100644 index 0000000000000000000000000000000000000000..9df32e9426614a1dc116dfc770bdca46f794c231 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/record.json @@ -0,0 +1,66 @@ +{ + "prompt_id": "multi_object_relations_d5_aac42d59f553", + "run_name": "flux2_qil_decomposition_pilot", + "category": "multi_object_relations", + "difficulty": 5, + "image_prompt": "Synthetic photorealistic scene with a bag on a shelf, a key inside the bag, and a notebook on top of the bag, high contrast and clear foreground separation.", + "target_evidence": "bag on shelf with key inside and notebook on top", + "expected_visual_elements": [ + "bag", + "shelf", + "key", + "notebook" + ], + "negative_prompt": "blurry, low contrast, overlapping objects, watermarks, real people, faces", + "diversity_tags": [ + "on top of", + "inside", + "shelf", + "bag" + ], + "llm_model_id": "Qwen/Qwen3-8B", + "prompt_hash": "aac42d59f553", + "rule_valid": true, + "rule_issues": [], + "rule_score": 1.0, + "opening_key": "synthetic photorealistic scene with a", + "max_similarity_to_other_prompt": 0.7391, + "diversity_score": 0.0609, + "repeated_opening": true, + "review": { + "prompt_id": "multi_object_relations_d5_aac42d59f553", + "accept": true, + "quality": 5, + "decomposition_fit": 5, + "diversity": 5, + "issues": [] + }, + "review_model_id": "/root/Layered-VQA/models/repos/Qwen_Qwen3-30B-A3B-Instruct-2507", + "final_rank_score": 1.0, + "status": "qwen_layered_done", + "image_model_id": "black-forest-labs/FLUX.2-dev", + "image_model_source": "/root/Layered-VQA/models/repos/black-forest-labs_FLUX.2-dev", + "image_model_slug": "black-forest-labs_FLUX.2-dev", + "image_path": "images/black-forest-labs_FLUX.2-dev/multi_object_relations/multi_object_relations_d5_aac42d59f553/image.png", + "image_width": 1024, + "image_height": 1024, + "num_inference_steps": 28, + "guidance_scale": 3.5, + "seed": 27540329, + "qwen_layered_model_id": "Qwen/Qwen-Image-Layered", + "qwen_layered_model_source": "/root/Layered-VQA/models/repos/Qwen_Qwen-Image-Layered", + "qwen_layered_prompt": "separate the image into clean transparent layers for foreground objects, readable text, occluders, relation-critical objects, and background", + "qwen_layer_count_requested": 3, + "qwen_layer_count_returned": 3, + "qwen_layer_paths": [ + "layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_00.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_01.png", + "layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/layers/layer_02.png" + ], + "qwen_resolution": 640, + "qwen_num_inference_steps": 5, + "qwen_true_cfg_scale": 4.0, + "qwen_seed": 1127660, + "qwen_elapsed_seconds": 5.207, + "qwen_record_path": "layers-x/k3/multi_object_relations/multi_object_relations_d5_aac42d59f553/record.json" +} diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_afeb8e5dfdd6/layers/layer_00.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_afeb8e5dfdd6/layers/layer_00.png new file mode 100644 index 0000000000000000000000000000000000000000..3e2723f6f5c2290a3b5cea55970302e3e696e5a9 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_afeb8e5dfdd6/layers/layer_00.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cda13c630e12b35c329c791905dcc400bf0af4f6e415cdb65fb828bc61cbe863 +size 458878 diff --git a/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_afeb8e5dfdd6/layers/layer_01.png b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_afeb8e5dfdd6/layers/layer_01.png new file mode 100644 index 0000000000000000000000000000000000000000..9240df5d01181bd0dd834470a18174b023160a50 --- /dev/null +++ b/vec_pilot/layers-x/k3/multi_object_relations/multi_object_relations_d5_afeb8e5dfdd6/layers/layer_01.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:32ef11ae6aac33044cca020d5f0a084e6dd289cdf5c429ed082cd9531293eb41 +size 544392