out t106b-F12_S06_s42 (9 files)

out/t106b-F12_S06_s42/F12_S06_s42.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2d33fbe6bb6a0f6de277eb1a46cfb670157774a4a87a30354a66e9540bb09da
+size 696989

out/t106b-F12_S06_s42/F12_S06_s42.prompt.txt

@@ -0,0 +1,12 @@
+subject_definitions:
+<Subject 1> is Boss, the black-and-white giant panda plush character; Boss's appearance comes from <Picture 1>. The description below calls <Subject 1> "Boss".
+
+retention_analysis:
+<Subject 1> (appears in [Shot 1]): fully_preserved - Boss keeps exactly the identity of <Picture 1> in every frame: boss, round black-and-white giant panda, ears with black fur, black eye patches, black limbs, white face and belly, no blush, no pink cheeks..
+
+detailed_description:
+[Shot 1] VISUAL RULE: no text, no subtitles, no captions, no letters. Dialogue is audio only. integrated_multimodal_description: At the start of the shot Boss points forward with one paw at chest height. By the end Boss holds the one-paw point, eyes on the camera. Boss is the only one talking: its tiny omega mouth opens on each syllable with a small pink tongue visible inside. boss, round black-and-white giant panda, ears with black fur, black eye patches, black limbs, white face and belly, no blush, no pink cheeks., in 高楼里的老板办公室，深色木桌，桌上有名牌和咖啡杯，落地窗外是城市。. The boss is in the center of the high-rise office, with a dark wooden desk, a nameplate, and a coffee cup, and a city view through the floor-to-ceiling window. The characters keep exactly the same 3D plush appearance in every frame: boss, round black-and-white giant panda, ears with black fur, black eye patches, black limbs, white face and belly, no blush, no pink cheeks., all with short stubby plush arms that stay close to the body and never reach, stretch or lengthen, and the same compact chibi head-to-body proportion in every shot. The black-and-white panda boss, with a calm, friendly adult male voice (S1), says: <d>[Chinese] 你以为五百万很多吗</d> Each character speaks exactly its own lines; the speaker tags are unique in this shot. Every character faces the camera enough for its face to be readable, unless the shot says otherwise; nobody turns their back to the camera. Camera: locked close-up, one character's head and shoulders filling the frame, no camera motion. Visual style: photoreal 3D plush film, velvety short plush fur, lighting matching the scene, film-grade color, no animation look, no cartoon outlines. The spoken lines are heard only in the audio and are never written on screen: no subtitles, no burned-in captions, no dialogue text, no speech bubbles. No text, captions, letters, signs, logos, watermark, extra characters, deformed hands, stretched, broken or bent-backwards limbs.
+
+overall_soundscape: quiet realistic indoor room tone, faint ambient hum, no music
+
+non_diegetic_music: N/A Wardrobe ownership is exact: boss: navy coat, navy trousers, yellow bow tie at the neck. Never transfer clothing or accessories between characters.

out/t106b-F12_S06_s42/result.json

@@ -0,0 +1,159 @@
+{
+ "unit_record": {
+  "id": "F12_S06_s42",
+  "status": "ok",
+  "mp4": "out/F12_S06_s42.mp4",
+  "frames": [
+   "out/F12_S06_s42_f000.png",
+   "out/F12_S06_s42_f011.png",
+   "out/F12_S06_s42_f022.png",
+   "out/F12_S06_s42_f033.png",
+   "out/F12_S06_s42_f044.png",
+   "out/F12_S06_s42_f055.png"
+  ],
+  "seed": 42,
+  "width": 832,
+  "height": 480,
+  "fps": 24,
+  "frames_rendered": 56,
+  "seconds": 2.333,
+  "lora": [
+   "t93m@0.8"
+  ],
+  "prompt_route": "ref_nots",
+  "gpu_seconds": 1465.0,
+  "error": null,
+  "rendered_at": "2026-10-07T16:11:30+0000",
+  "warnings": [],
+  "timestamps_stripped": 2,
+  "audio_lock": true,
+  "graphs": [
+   {
+    "tag": "F12_S06_s42_enc",
+    "ok": true,
+    "sec": 1026.7,
+    "warn": []
+   },
+   {
+    "tag": "F12_S06_s42_smp",
+    "ok": true,
+    "sec": 438.3,
+    "warn": []
+   }
+  ],
+  "wall_seconds": 1518.6,
+  "peak_vram_mib": 13019,
+  "min_ram_avail_mib": 20607,
+  "cgroup_peak_mib": 30706,
+  "media": {
+   "path": "F12_S06_s42.mp4",
+   "streams": [
+    {
+     "codec_name": "h264",
+     "codec_type": "video",
+     "width": 832,
+     "height": 480,
+     "duration": "2.333333",
+     "nb_read_frames": "56"
+    },
+    {
+     "codec_name": "aac",
+     "codec_type": "audio",
+     "sample_rate": "32000",
+     "channels": 2,
+     "duration": "2.325000",
+     "nb_read_frames": "73"
+    }
+   ],
+   "width": 832,
+   "height": 480,
+   "frames": 56,
+   "duration": 2.333333,
+   "luma_mean": 129.45,
+   "black_frames": 0,
+   "temporal_diff_mean": 2.57,
+   "has_audio": true
+  }
+ },
+ "batch_results": {
+  "kernel": {
+   "owner": "moonberry53",
+   "slug": "t106b-3",
+   "gpu": null,
+   "started": "2026-10-07T15:46:11+0000",
+   "ended": null
+  },
+  "totals": {
+   "units_requested": 1,
+   "units_ok": 0,
+   "units_error": 0,
+   "units_skipped": 0,
+   "gpu_seconds": 0.0,
+   "peak_vram_mib": 0,
+   "min_ram_avail_mib": null,
+   "cgroup_peak_mib": 0,
+   "setup_seconds": null,
+   "running": "F12_S06_s42",
+   "gpu_seconds_so_far": 1465.0
+  },
+  "config": {
+   "width": 608,
+   "height": 352,
+   "frames": 124,
+   "fps": 24,
+   "prompt_route": "text",
+   "steps": 4,
+   "turbo_scale": 1.0,
+   "shift_video": 6.0,
+   "shift_audio": 3.0,
+   "fp16safe": true,
+   "reserve_vram": 2.4,
+   "graph_timeout": 3600,
+   "ballast_gb": 2.0,
+   "te_device": "cuda:1",
+   "max_min": 420,
+   "min_left_s": 300,
+   "comfy_ref": "v0.37.0",
+   "comfy_ref_audio_lock": "v0.37.2",
+   "weights_dataset": "adamzale/minimax-h3-fl2va-weights",
+   "dit": "minimax_h3_ref2va_pruned_fp8_scaled.safetensors",
+   "refmod_dataset": "",
+   "turbo": "minimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensors",
+   "sampler": "res_multistep",
+   "ref_image_size": "match",
+   "decode_fast": true,
+   "decode_tile_size": 512,
+   "audio_start_frame": 6,
+   "audio_gain_db": 0.0,
+   "owner": "moonberry53",
+   "slug": "t106b-3",
+   "batch": "t106b",
+   "gpu": "",
+   "dry_run": false,
+   "units_path": null,
+   "input_dir": "/kaggle/working/g0/input",
+   "work_dir": "/kaggle/working/g0",
+   "tmp_dir": "/kaggle/temp/g0",
+   "out_dir": "/kaggle/working/g0/out",
+   "payload_dataset": null,
+   "extra_loras": [],
+   "version": "1.0.0"
+  },
+  "lora": [
+   {
+    "name": "t93m",
+    "source": "payload:loras/t93m-e0300.safetensors",
+    "scale": 0.8,
+    "kind": "payload",
+    "file_name": "t93m-e0300_comfy.safetensors",
+    "label": "t93m@0.8"
+   }
+  ],
+  "units": []
+ },
+ "gpu": 0,
+ "renderer_setup_s": 839.8,
+ "assets_s": 17.6,
+ "wall_s": 1518.6,
+ "comfy_port": 8188
+}