Add files using upload-large-folder tool
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_000.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_001.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_003.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_004.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_005.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_006.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_008.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_009.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_010.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_011.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_012.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_013.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_014.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_015.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_016.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_017.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_018.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_019.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_020.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_021.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_022.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_024.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_025.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_027.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_028.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_029.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_031.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_032.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_034.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_035.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_036.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_038.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_039.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_041.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_042.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_043.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_044.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_045.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_047.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_048.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_050.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_052.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_053.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_054.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_058.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_059.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_061.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_062.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_063.json +38 -0
- eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_064.json +38 -0
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_000.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 683,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 0,
|
| 13 |
+
"prompt": "A close-up view of a cluster of vibrant green grapes on a rotating glass table under soft, diffused lighting. The grapes are large and plump, reflecting the gentle light as they rotate slowly, showcasing their smooth, shiny surfaces. The background is blurred, focusing attention solely on the grapes and their subtle reflections. The camera remains static, capturing the serene and detailed motion of the rotating grapes. Close-up shot.",
|
| 14 |
+
"original_prompt": "Close up of grapes on a rotating table.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/000.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2694.834171295166,
|
| 22 |
+
"excluded_context_kv_latency_ms": 783.2665557861328,
|
| 23 |
+
"wall_generation_s": 5.235450267791748,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.07594413310289383,
|
| 27 |
+
"psnr": 11.19505763053894,
|
| 28 |
+
"ssim": 0.499019056558609,
|
| 29 |
+
"lpips": 0.5706181526184082,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.464102564102564,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.464102564102564
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_001.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 684,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 1,
|
| 13 |
+
"prompt": "A single green sea turtle gracefully swimming through the vast, clear waters of the ocean. The turtle has a smooth, rounded shell with brown and olive patterns, and flippers that move rhythmically as it propels itself forward. Schools of colorful fish swim alongside, adding to the vibrant marine life. The sunlight filters through the water, creating gentle beams and highlighting the soft, turquoise hues of the ocean. The background shows a few coral reefs and patches of seaweed swaying gently with the current. Shot from an underwater perspective, medium close-up focusing on the turtle.",
|
| 14 |
+
"original_prompt": "Turtle swimming in ocean.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/001.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2809.2373809814453,
|
| 22 |
+
"excluded_context_kv_latency_ms": 793.4015579223633,
|
| 23 |
+
"wall_generation_s": 5.36639928817749,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.028700893744826317,
|
| 27 |
+
"psnr": 15.421046018600464,
|
| 28 |
+
"ssim": 0.5757187604904175,
|
| 29 |
+
"lpips": 0.39291882514953613,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.292948717948715,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.292948717948717
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_003.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 686,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 3,
|
| 13 |
+
"prompt": "A large, black-and-white panda stands confidently on a surfboard in the golden hour of sunset over the ocean. The panda has a playful expression, with its front paws resting lightly on the board. The waves gently roll around the surfboard, and the sky is painted with warm hues of orange and pink. The water is calm, reflecting the vibrant colors of the setting sun. The scene captures the serene beauty of the ocean at twilight. Medium shot focusing on the panda and the surfboard, with the horizon and sunset visible in the background.",
|
| 14 |
+
"original_prompt": "A panda standing on a surfboard in the ocean in sunset.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/003.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2792.150260925293,
|
| 22 |
+
"excluded_context_kv_latency_ms": 788.7437133789062,
|
| 23 |
+
"wall_generation_s": 5.339996814727783,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.03104514814913273,
|
| 27 |
+
"psnr": 15.0800621509552,
|
| 28 |
+
"ssim": 0.6508263349533081,
|
| 29 |
+
"lpips": 0.3930548131465912,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.627136752136753,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.627136752136753
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_004.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 687,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 4,
|
| 13 |
+
"prompt": "A serene sunny afternoon scene featuring an astronaut in a sleek, white spacesuit with a reflective visor, feeding ducks at a tranquil pond. The astronaut bends gently at the waist, holding a handful of bread crumbs towards the ducks, which are swimming close to the surface. The water reflects the clear blue sky and the astronaut's figure, creating a harmonious mirror image. The surrounding environment includes lush green grass and blooming flowers, enhancing the peaceful atmosphere. The camera focuses closely on the interaction between the astronaut and the ducks, capturing the gentle movements and reflections in the water. Medium close-up shot, static camera.",
|
| 14 |
+
"original_prompt": "An astronaut feeding ducks on a sunny afternoon, reflection from the water.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/004.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2758.7766151428223,
|
| 22 |
+
"excluded_context_kv_latency_ms": 783.9827499389648,
|
| 23 |
+
"wall_generation_s": 5.300414085388184,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.06209418177604675,
|
| 27 |
+
"psnr": 12.069491147994995,
|
| 28 |
+
"ssim": 0.3278732895851135,
|
| 29 |
+
"lpips": 0.45329639315605164,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.46944444444445,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.469444444444445
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_005.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 688,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 5,
|
| 13 |
+
"prompt": "A cozy indoor setting with two pandas sitting on plush chairs, engaged in a thoughtful discussion about an academic paper. Both pandas have black patches around their eyes and are wearing small glasses perched on their snouts. They are hunched over a stack of papers filled with complex equations and graphs, occasionally pointing at sections and nodding in agreement. One panda is gesturing animatedly with a pen, while the other listens intently. The room is softly lit with warm, ambient lighting, creating a comfortable atmosphere for scholarly debate. Medium close-up view focusing on the pandas' expressions and gestures.",
|
| 14 |
+
"original_prompt": "Two pandas discussing an academic paper.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/005.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2791.6986083984375,
|
| 22 |
+
"excluded_context_kv_latency_ms": 788.8612060546875,
|
| 23 |
+
"wall_generation_s": 5.339401721954346,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.05427330732345581,
|
| 27 |
+
"psnr": 12.65413761138916,
|
| 28 |
+
"ssim": 0.46675384044647217,
|
| 29 |
+
"lpips": 0.41816022992134094,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.53653846153846,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.536538461538462
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_006.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 689,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 6,
|
| 13 |
+
"prompt": "A sunset time-lapse sequence at a serene beach, capturing the vibrant colors of the sky as the sun sets below the horizon. The sky transitions from bright orange and pink hues to deep purples and blues, with wispy clouds moving gracefully across the frame. The beach is calm, with soft waves gently lapping at the shore. The camera remains stationary, focusing on the dynamic changes in the sky and the interplay of light and shadow over the landscape. Wide shot, capturing the expansive beauty of the beach and sky during sunset.",
|
| 14 |
+
"original_prompt": "Sunset time lapse at the beach with moving clouds and colors in the sky.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/006.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2746.414306640625,
|
| 22 |
+
"excluded_context_kv_latency_ms": 787.4128112792969,
|
| 23 |
+
"wall_generation_s": 5.293547630310059,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.020054765045642853,
|
| 27 |
+
"psnr": 16.977823972702026,
|
| 28 |
+
"ssim": 0.6060497164726257,
|
| 29 |
+
"lpips": 0.3720584809780121,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.455555555555552,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.455555555555555
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_008.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 691,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 8,
|
| 13 |
+
"prompt": "A koala bear playing a grand piano in a lush, dense forest. The koala has soft, grey fur and large, round ears. It sits upright on the piano bench, paws delicately placed on the keys, creating gentle melodies. The forest background is filled with tall eucalyptus trees, dappled sunlight filtering through the leaves, and a carpet of green moss beneath the piano. The scene is calm and serene, with the koala focused intently on its performance. Medium close-up shot, capturing the koala and part of the forest surroundings.",
|
| 14 |
+
"original_prompt": "A koala bear playing piano in the forest.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/008.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2783.821647644043,
|
| 22 |
+
"excluded_context_kv_latency_ms": 785.6505355834961,
|
| 23 |
+
"wall_generation_s": 5.328362226486206,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.04621875286102295,
|
| 27 |
+
"psnr": 13.351818323135376,
|
| 28 |
+
"ssim": 0.2623670697212219,
|
| 29 |
+
"lpips": 0.4381442070007324,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.52777777777778,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.527777777777778
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_009.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 692,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 9,
|
| 13 |
+
"prompt": "A realistic animation of an astronaut floating in the vastness of space. The astronaut is wearing a full spacesuit with reflective gold visor and carrying a jetpack. They are gliding gracefully, arms outstretched, against a backdrop of stars and distant planets. The scene includes streaks of light from passing meteoroids and a faint blue Earth visible in the distance. The astronaut occasionally adjusts their jetpack to maneuver smoothly through space. Wide shot, static camera, emphasizing the serene beauty of the cosmos.",
|
| 14 |
+
"original_prompt": "An astronaut flying in space.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/009.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2803.463523864746,
|
| 22 |
+
"excluded_context_kv_latency_ms": 791.5273971557617,
|
| 23 |
+
"wall_generation_s": 5.357872486114502,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.032588355243206024,
|
| 27 |
+
"psnr": 14.869375228881836,
|
| 28 |
+
"ssim": 0.5544800162315369,
|
| 29 |
+
"lpips": 0.39670830965042114,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.41559829059829,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.4155982905982905
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_010.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 693,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 10,
|
| 13 |
+
"prompt": "A vibrant display of fireworks lighting up the night sky. The fireworks burst into colorful explosions of red, blue, green, and gold, creating intricate patterns against a dark, starry backdrop. Each explosion is followed by trails of light that slowly fade away. The ground is seen below, with people watching in awe, their faces illuminated momentarily by the bright flashes. The scene captures the excitement and joy of a festive celebration. Wide shot, capturing the expansive sky and the crowd below.",
|
| 14 |
+
"original_prompt": "Fireworks.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/010.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2710.5640258789062,
|
| 22 |
+
"excluded_context_kv_latency_ms": 787.0962600708008,
|
| 23 |
+
"wall_generation_s": 5.252301931381226,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.04941364750266075,
|
| 27 |
+
"psnr": 13.061530590057373,
|
| 28 |
+
"ssim": 0.35586896538734436,
|
| 29 |
+
"lpips": 0.48954838514328003,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.59380341880342,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.593803418803417
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_011.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 694,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 11,
|
| 13 |
+
"prompt": "Animated watercolor painting, soft and fluffy white clouds drifting across a bright blue sky. The clouds move gracefully, changing shape as they float, creating a serene and tranquil atmosphere. The sky is clear with gentle sunlight filtering through, casting a warm glow over the scene. The clouds are rendered with delicate brush strokes, giving them a dreamy, almost ethereal quality. Close-up view focusing on the movement and transformation of the clouds.",
|
| 14 |
+
"original_prompt": "An animated painting of fluffy white clouds moving in sky.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/011.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2759.5183715820312,
|
| 22 |
+
"excluded_context_kv_latency_ms": 789.8494415283203,
|
| 23 |
+
"wall_generation_s": 5.310692548751831,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.026202267035841942,
|
| 27 |
+
"psnr": 15.816611051559448,
|
| 28 |
+
"ssim": 0.815554141998291,
|
| 29 |
+
"lpips": 0.41166412830352783,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.2508547008547,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.250854700854703
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_012.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 695,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 12,
|
| 13 |
+
"prompt": "Aerial view flying through vibrant fantasy landscapes. The camera captures sweeping vistas of floating islands, lush forests, and majestic waterfalls. Enormous mythical creatures like dragons and unicorns roam freely in the distance. The sky transitions from dawn hues to midday sunlight, casting dynamic shadows on the terrain. The landscape is detailed with intricate flora and fauna, creating a rich and immersive environment. The camera moves smoothly, capturing the vastness and beauty of these magical realms. Wide shot aerial perspective throughout the sequence.",
|
| 14 |
+
"original_prompt": "Flying through fantasy landscapes.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/012.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2763.3977966308594,
|
| 22 |
+
"excluded_context_kv_latency_ms": 782.9016723632812,
|
| 23 |
+
"wall_generation_s": 5.305049896240234,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.039403609931468964,
|
| 27 |
+
"psnr": 14.044640064239502,
|
| 28 |
+
"ssim": 0.47450515627861023,
|
| 29 |
+
"lpips": 0.4843088388442993,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.410470085470088,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.410470085470085
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_013.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 696,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 13,
|
| 13 |
+
"prompt": "A large, hairy Bigfoot creature walking through a heavy snowstorm. The Bigfoot stands at least eight feet tall, covered in shaggy brown fur, with muscular limbs and a stooped posture. Snowflakes swirl around him as he moves slowly and deliberately through the dense forest, his feet sinking slightly into the deep snow. The landscape is bleak and desolate, with bare trees and thick snowdrifts. The atmosphere is eerie and quiet, with only the sound of crunching snow and howling wind. The scene is captured in a mid-shot, focusing on the Bigfoot\u2019s powerful form as he trudges through the storm.",
|
| 14 |
+
"original_prompt": "A bigfoot walking in the snowstorm.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/013.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2780.117401123047,
|
| 22 |
+
"excluded_context_kv_latency_ms": 787.3561859130859,
|
| 23 |
+
"wall_generation_s": 5.329988241195679,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.039699897170066833,
|
| 27 |
+
"psnr": 14.012106657028198,
|
| 28 |
+
"ssim": 0.36132606863975525,
|
| 29 |
+
"lpips": 0.49621760845184326,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.26346153846154,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.263461538461539
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_014.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 697,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 14,
|
| 13 |
+
"prompt": "A realistic animation of a cute, bushy-tailed squirrel sitting on a tree branch, holding a juicy burger in its tiny paws. The squirrel has a curious expression as it takes a bite out of the burger, revealing its sharp little teeth. Its fur is brown with lighter underparts, and its large, expressive eyes convey excitement and delight. The background shows a lush forest with sunlight filtering through the leaves, casting dappled shadows. The squirrel is focused on the burger, ignoring its surroundings. Medium close-up shot to capture the squirrel's facial expressions and the burger clearly.",
|
| 14 |
+
"original_prompt": "A squirrel eating a burger.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/014.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2805.314785003662,
|
| 22 |
+
"excluded_context_kv_latency_ms": 787.6884155273438,
|
| 23 |
+
"wall_generation_s": 5.355424404144287,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.04496455937623978,
|
| 27 |
+
"psnr": 13.471297025680542,
|
| 28 |
+
"ssim": 0.41380056738853455,
|
| 29 |
+
"lpips": 0.32745569944381714,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.61709401709402,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.617094017094017
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_015.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 698,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 15,
|
| 13 |
+
"prompt": "A sleek black cat wearing stylish oversized sunglasses and a bright red lifeguard hat, positioned beside a pool. The cat is sitting upright with its paws resting on the edge of a poolside chair, alert and focused. The cat's fur is smooth and glossy, with a collar that has a small whistle attached. The pool is clear and sparkling, surrounded by lush green grass and palm trees. In the background, there are lounge chairs and umbrellas under a bright blue sky. The scene is set in a warm, sunny day environment. Medium close-up shot, static camera.",
|
| 14 |
+
"original_prompt": "A cat wearing sunglasses and working as a lifeguard at a pool.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/015.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2739.323329925537,
|
| 22 |
+
"excluded_context_kv_latency_ms": 782.3462677001953,
|
| 23 |
+
"wall_generation_s": 5.2820823192596436,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.06230203062295914,
|
| 27 |
+
"psnr": 12.054977416992188,
|
| 28 |
+
"ssim": 0.43182727694511414,
|
| 29 |
+
"lpips": 0.4298038184642792,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.548931623931622,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.548931623931625
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_016.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 699,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 16,
|
| 13 |
+
"prompt": "A sweeping panoramic view of snow-covered rocky mountain peaks surrounding deep, shadowed canyons. The canyons twist and wind through the high-elevated terrain, showcasing dramatic vistas and textured rock formations. Snow blankets the jagged peaks, creating a serene yet awe-inspiring winter landscape. The camera gradually pans across the vast expanse, emphasizing the scale and beauty of the natural environment. Wide shot, capturing the grandeur of the mountains and canyons.",
|
| 14 |
+
"original_prompt": "Snow rocky mountains peaks canyon. snow blanketed rocky mountains surround and shadow deep canyons. the canyons twist and bend through the high elevated mountain peaks.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/016.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2721.6052856445312,
|
| 22 |
+
"excluded_context_kv_latency_ms": 786.2397994995117,
|
| 23 |
+
"wall_generation_s": 5.266711473464966,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.04699341952800751,
|
| 27 |
+
"psnr": 13.279629945755005,
|
| 28 |
+
"ssim": 0.42430511116981506,
|
| 29 |
+
"lpips": 0.376659095287323,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.477350427350423,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.477350427350428
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_017.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 700,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 17,
|
| 13 |
+
"prompt": "Extreme slow-motion video of a splash of turquoise water, showcasing intricate droplets and ripples forming in vivid detail. The scene focuses on the moment just after impact, capturing the fluid dynamics and transparency of the water. The alpha channel is included, allowing for a seamless blend with any background. Shot from a close-up perspective to highlight the fine textures and movement of the water.",
|
| 14 |
+
"original_prompt": "Splash of turquoise water in extreme slow motion, alpha channel included.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/017.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2835.8428230285645,
|
| 22 |
+
"excluded_context_kv_latency_ms": 790.925895690918,
|
| 23 |
+
"wall_generation_s": 5.3948609828948975,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.05113092064857483,
|
| 27 |
+
"psnr": 12.913163900375366,
|
| 28 |
+
"ssim": 0.41156288981437683,
|
| 29 |
+
"lpips": 0.5442873239517212,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.23183760683761,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.231837606837607
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_018.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 701,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 18,
|
| 13 |
+
"prompt": "A single scoop of vanilla ice cream is slowly melting on a wooden table. The ice cream has a glossy surface with small drips forming at the edges, creating small pools of melted ice cream on the table. The wooden table has a warm, natural finish with visible grain patterns. The background includes blurred elements of a kitchen, such as a countertop and appliances, giving context to the scene. The ice cream melts gradually, with the camera remaining static to capture the process. Close-up shot focusing on the melting ice cream.",
|
| 14 |
+
"original_prompt": "an ice cream is melting on the table.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/018.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2788.4016151428223,
|
| 22 |
+
"excluded_context_kv_latency_ms": 786.1833343505859,
|
| 23 |
+
"wall_generation_s": 5.329174995422363,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.015445472672581673,
|
| 27 |
+
"psnr": 18.111988306045532,
|
| 28 |
+
"ssim": 0.638966977596283,
|
| 29 |
+
"lpips": 0.353475421667099,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.132051282051282,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.132051282051282
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_019.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 702,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 19,
|
| 13 |
+
"prompt": "A drone flying smoothly over a snowy forest, capturing aerial footage from a high vantage point. The forest is covered in a thick blanket of snow, with tall pine trees reaching towards the bright winter sky. The branches are laden with snow, and the ground is pristine and untouched. The camera slowly pans across the landscape, revealing the serene beauty of the snowy expanse. The scene is bathed in soft, diffused light, creating a tranquil and peaceful atmosphere. Wide shot, aerial view.",
|
| 14 |
+
"original_prompt": "a drone flying over a snowy forest.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/019.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2778.063465118408,
|
| 22 |
+
"excluded_context_kv_latency_ms": 788.3982620239258,
|
| 23 |
+
"wall_generation_s": 5.326310634613037,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.06552514433860779,
|
| 27 |
+
"psnr": 11.835919618606567,
|
| 28 |
+
"ssim": 0.218252494931221,
|
| 29 |
+
"lpips": 0.5146428346633911,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.179487179487186,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.17948717948718
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_020.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 703,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 20,
|
| 13 |
+
"prompt": "A large great white shark is swimming gracefully through the vast, deep blue ocean. Its sleek, muscular body moves powerfully through the water, propelling it forward with each effortless stroke of its tail. The shark's distinctive dorsal fin cuts through the surface as it glides past schools of smaller fish, which scatter in its wake. The ocean floor is visible below, covered in soft sand and dotted with coral formations. The water is clear, allowing sunlight to penetrate and create shimmering patterns around the shark. The scene is captured in a mid-shot, maintaining a steady camera position to showcase the shark's majestic presence and the serene underwater environment.",
|
| 14 |
+
"original_prompt": "a shark is swimming in the ocean.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/020.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2715.4427185058594,
|
| 22 |
+
"excluded_context_kv_latency_ms": 782.3945693969727,
|
| 23 |
+
"wall_generation_s": 5.25596809387207,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.03399492800235748,
|
| 27 |
+
"psnr": 14.685858488082886,
|
| 28 |
+
"ssim": 0.5878732800483704,
|
| 29 |
+
"lpips": 0.3829149305820465,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.412820512820513,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.412820512820512
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_021.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 704,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 21,
|
| 13 |
+
"prompt": "An aerial panoramic video captured from a drone flying over a fantastical landscape. The scene showcases a vibrant, otherworldly terrain with towering mystical castles, lush enchanted forests, and shimmering lakes surrounded by floating islands. The sky is filled with colorful clouds and mythical creatures flying gracefully. The camera moves smoothly in a wide, sweeping motion, capturing the grandeur of this magical realm from various angles and elevations. The video maintains a continuous panoramic view, emphasizing the vastness and wonder of the fantasy land.",
|
| 14 |
+
"original_prompt": "Aerial panoramic video from a drone of a fantasy land.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/021.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2717.9471321105957,
|
| 22 |
+
"excluded_context_kv_latency_ms": 787.5477981567383,
|
| 23 |
+
"wall_generation_s": 5.265438556671143,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.04314759373664856,
|
| 27 |
+
"psnr": 13.650434017181396,
|
| 28 |
+
"ssim": 0.450022429227829,
|
| 29 |
+
"lpips": 0.4980117380619049,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.391880341880345,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.391880341880341
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_022.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 705,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 22,
|
| 13 |
+
"prompt": "A cute, plush teddy bear is swimming in the vast ocean. The teddy bear, with its soft brown fur and big, round black buttons for eyes, is floating on the water's surface. It has a small red bow tie around its neck. The waves gently rock the teddy bear as it bobs up and down. In the background, there are rolling waves and a bright blue sky with fluffy white clouds. The water sparkles under the sunlight. The scene is captured from a mid-shot perspective, showing the teddy bear and a portion of the ocean surrounding it.",
|
| 14 |
+
"original_prompt": "a teddy bear is swimming in the ocean.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/022.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2725.1075477600098,
|
| 22 |
+
"excluded_context_kv_latency_ms": 789.1935653686523,
|
| 23 |
+
"wall_generation_s": 5.2740561962127686,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.04431906342506409,
|
| 27 |
+
"psnr": 13.534094095230103,
|
| 28 |
+
"ssim": 0.557174026966095,
|
| 29 |
+
"lpips": 0.36527934670448303,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.713461538461537,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.713461538461537
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_024.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 707,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 24,
|
| 13 |
+
"prompt": "A vibrant scene of golden fish swimming gracefully in an expansive underwater ocean. The fish shimmer with metallic gold scales, their bodies undulating as they navigate through crystal-clear water. Schools of these fish move together, creating a mesmerizing pattern of synchronized swimming. The background showcases a diverse marine environment with coral reefs, sea plants, and other colorful marine life. The camera remains static, capturing the fluid motion and beauty of the golden fish in a wide-angle shot, emphasizing the vastness of the ocean.",
|
| 14 |
+
"original_prompt": "golden fish swimming in the ocean.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/024.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2773.7555923461914,
|
| 22 |
+
"excluded_context_kv_latency_ms": 787.9650192260742,
|
| 23 |
+
"wall_generation_s": 5.3221094608306885,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.08281075209379196,
|
| 27 |
+
"psnr": 10.819132328033447,
|
| 28 |
+
"ssim": 0.36432647705078125,
|
| 29 |
+
"lpips": 0.5390549302101135,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.478846153846156,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.478846153846154
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_025.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 708,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 25,
|
| 13 |
+
"prompt": "Close-up of an artist painting on a canvas using a large round brush. The artist has medium-length brown hair tied back in a loose ponytail and wears a white apron over their clothes. They are focused intently, their hand moving smoothly across the canvas. The brushstrokes are visible, blending colors together in fluid motions. The background shows part of the artist's workspace, with other brushes and paint tubes nearby. The scene emphasizes the intricate detail and the flowing motion of the brush as it interacts with the paint and canvas.",
|
| 14 |
+
"original_prompt": "An artist brush painting on a canvas close up.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/025.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2719.4487648010254,
|
| 22 |
+
"excluded_context_kv_latency_ms": 787.7677383422852,
|
| 23 |
+
"wall_generation_s": 5.2698423862457275,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.03864230588078499,
|
| 27 |
+
"psnr": 14.129369258880615,
|
| 28 |
+
"ssim": 0.5425694584846497,
|
| 29 |
+
"lpips": 0.49952998757362366,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.402136752136755,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.4021367521367525
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_027.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 710,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 27,
|
| 13 |
+
"prompt": "A happy, medium-sized dog wearing a bright yellow turtleneck sweater sits in a studio setting. The dog has a friendly expression with its tail wagging gently and ears perked up, facing the camera directly. It is standing in a relaxed posture with its front paws slightly apart. The background is entirely dark, creating a stark contrast with the dog\u2019s cheerful demeanor. The lighting highlights the dog\u2019s fur and the texture of the turtleneck. Close-up portrait shot.",
|
| 14 |
+
"original_prompt": "happy dog wearing a yellow turtleneck, studio, portrait, facing camera, dark background",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/027.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2809.2482872009277,
|
| 22 |
+
"excluded_context_kv_latency_ms": 789.9344100952148,
|
| 23 |
+
"wall_generation_s": 5.3605523109436035,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.016055909916758537,
|
| 27 |
+
"psnr": 17.943650484085083,
|
| 28 |
+
"ssim": 0.7755357027053833,
|
| 29 |
+
"lpips": 0.14620789885520935,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.320512820512818,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.320512820512821
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_028.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 711,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 28,
|
| 13 |
+
"prompt": "3D render of origami dancers made from white paper, performing a modern dance routine against a pristine white background in a studio setting. Each dancer has delicate folds and creases that catch the light, enhancing their ethereal appearance. They are gracefully moving in synchronized motions, expressing fluidity and elegance through their postures. The scene focuses on a medium close-up to capture the intricate details of the dancers' movements and the soft, clean background.",
|
| 14 |
+
"original_prompt": "Origami dancers in white paper, 3D render, on white background, studio shot, dancing modern dance.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/028.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2788.4898223876953,
|
| 22 |
+
"excluded_context_kv_latency_ms": 784.8173522949219,
|
| 23 |
+
"wall_generation_s": 5.332075357437134,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.005380737129598856,
|
| 27 |
+
"psnr": 22.69158124923706,
|
| 28 |
+
"ssim": 0.8290687799453735,
|
| 29 |
+
"lpips": 0.3426859378814697,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.65833333333333,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.658333333333334
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_029.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 712,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 29,
|
| 13 |
+
"prompt": "A cozy campfire scene at night in a snowy forest, with a bright, warm fire casting flickering light and shadows. The ground is covered in pristine snow, and tall pine trees surround the area, their branches heavy with snow. The sky above is clear and starlit, with twinkling stars visible overhead. A few logs crackle and pop in the fire, and smoke gently rises into the cool night air. The overall atmosphere is serene and peaceful, with a sense of warmth amidst the cold surroundings. Wide shot, capturing the expansive snowy landscape and the inviting glow of the campfire.",
|
| 14 |
+
"original_prompt": "Campfire at night in a snowy forest with starry sky in the background.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/029.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2793.162078857422,
|
| 22 |
+
"excluded_context_kv_latency_ms": 788.3548889160156,
|
| 23 |
+
"wall_generation_s": 5.341973543167114,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.019865019246935844,
|
| 27 |
+
"psnr": 17.01910972595215,
|
| 28 |
+
"ssim": 0.45479556918144226,
|
| 29 |
+
"lpips": 0.3565388321876526,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.32905982905983,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.329059829059829
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_031.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 714,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 31,
|
| 13 |
+
"prompt": "A detailed 3D model of a grand Victorian house from the 1800s, showcasing intricate architectural details such as ornate railings, multiple gables, and large bay windows. The house is painted in classic muted tones with a white trim, and surrounded by a well-manicured lawn and flower beds. The exterior features include a wrap-around porch with elegant columns and hanging lanterns. The scene captures the essence of late 19th-century elegance, with a serene and picturesque atmosphere. Wide shot, static scene.",
|
| 14 |
+
"original_prompt": "A 3D model of a 1800s victorian house.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/031.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2784.4314460754395,
|
| 22 |
+
"excluded_context_kv_latency_ms": 786.1396484375,
|
| 23 |
+
"wall_generation_s": 5.328782320022583,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.048362888395786285,
|
| 27 |
+
"psnr": 13.154877424240112,
|
| 28 |
+
"ssim": 0.3537846803665161,
|
| 29 |
+
"lpips": 0.3627426028251648,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.414957264957266,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.414957264957265
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_032.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 715,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 32,
|
| 13 |
+
"prompt": "A first-person perspective video showing the process of applying makeup in the morning. The video captures detailed close-ups of each step, such as applying foundation, concealer, blush, eyeshadow, eyeliner, mascara, and lipstick. The makeup artist is seen using various brushes and tools with precise motions. The lighting is soft and flattering, highlighting the natural glow of the skin. The camera remains static during each step, focusing solely on the application process. The overall style is tutorial-like, providing clear instructions and tips throughout the video.",
|
| 14 |
+
"original_prompt": "this is how I do makeup in the morning.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/032.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2736.740074157715,
|
| 22 |
+
"excluded_context_kv_latency_ms": 783.8841247558594,
|
| 23 |
+
"wall_generation_s": 5.277477979660034,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.03136055916547775,
|
| 27 |
+
"psnr": 15.03616213798523,
|
| 28 |
+
"ssim": 0.5859060883522034,
|
| 29 |
+
"lpips": 0.30810317397117615,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.379700854700857,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.379700854700856
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_034.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 717,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 34,
|
| 13 |
+
"prompt": "A futuristic robot performing an energetic dance routine in the bustling heart of Times Square. The robot, with sleek metallic surfaces and glowing LED lights, moves gracefully among the neon signs and crowds of people. It performs a variety of fluid, robotic movements, including spins, arm waves, and foot taps. The background showcases iconic billboards and the vibrant nightlife of Times Square at night, with bright lights and moving advertisements. The scene is captured from a mid-shot perspective, focusing on the robot's expressive dance, with no camera movement.",
|
| 14 |
+
"original_prompt": "Robot dancing in Times Square.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/034.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2774.316883087158,
|
| 22 |
+
"excluded_context_kv_latency_ms": 787.7195434570312,
|
| 23 |
+
"wall_generation_s": 5.315627098083496,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.10932675004005432,
|
| 27 |
+
"psnr": 9.612735509872437,
|
| 28 |
+
"ssim": 0.28149789571762085,
|
| 29 |
+
"lpips": 0.48435041308403015,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.233760683760686,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.233760683760684
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_035.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 718,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 35,
|
| 13 |
+
"prompt": "A busy freeway at night, showcasing multiple lanes filled with cars moving at varying speeds. The scene is illuminated by the bright headlights and taillights of the vehicles, casting reflections on the wet asphalt. The environment is bustling with activity, featuring several types of vehicles including sedans, SUVs, trucks, and motorcycles. The distant horizon is blurred by the lights from oncoming traffic, creating a vibrant and dynamic scene. The background includes dimly lit streetlights and occasional billboards. The camera captures the motion of the vehicles in a wide shot, maintaining a stable perspective to highlight the flow of traffic.",
|
| 14 |
+
"original_prompt": "Busy freeway at night.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/035.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2736.6767501831055,
|
| 22 |
+
"excluded_context_kv_latency_ms": 791.9557723999023,
|
| 23 |
+
"wall_generation_s": 5.2890565395355225,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.04545409604907036,
|
| 27 |
+
"psnr": 13.424270153045654,
|
| 28 |
+
"ssim": 0.4079713225364685,
|
| 29 |
+
"lpips": 0.49969610571861267,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.549999999999997,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.549999999999998
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_036.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 719,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 36,
|
| 13 |
+
"prompt": "In extreme slow motion, a transparent balloon filled with water hangs suspended in mid-air. The balloon gradually stretches as the tension builds, revealing droplets of water clinging to its surface. Suddenly, the balloon bursts, releasing a shower of water droplets that scatter and splatter in all directions. The camera remains static, capturing every intricate detail of the explosion in breathtaking slow motion, showcasing the fluid dynamics and the vivid splash of water. Close-up shot focusing on the balloon before and during the explosion.",
|
| 14 |
+
"original_prompt": "Balloon full of water exploding in extreme slow motion.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/036.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2784.7388229370117,
|
| 22 |
+
"excluded_context_kv_latency_ms": 785.1405868530273,
|
| 23 |
+
"wall_generation_s": 5.32740592956543,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.0456833653151989,
|
| 27 |
+
"psnr": 13.402419090270996,
|
| 28 |
+
"ssim": 0.6273183226585388,
|
| 29 |
+
"lpips": 0.3872758448123932,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.335042735042737,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.335042735042735
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_038.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 721,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 38,
|
| 13 |
+
"prompt": "Macro slow-motion video. A cropped close-up of roasted coffee beans falling gracefully into an empty bowl. The beans are glossy with rich brown hues and intricate textures. As they descend, the beans emit a soft rustling sound, emphasizing their natural beauty and the smooth motion of their fall. The bowl is plain and white, providing a stark contrast to the dark beans. The camera focuses tightly on each bean as it arcs through the air before landing softly in the bowl. The scene is captured in slow motion to highlight the detailed movement and the sensory experience of the falling beans.",
|
| 14 |
+
"original_prompt": "Macro slo-mo. Slow motion cropped closeup of roasted coffee beans falling into an empty bowl.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/038.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2723.1055641174316,
|
| 22 |
+
"excluded_context_kv_latency_ms": 786.6792907714844,
|
| 23 |
+
"wall_generation_s": 5.269315719604492,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.07944395393133163,
|
| 27 |
+
"psnr": 10.999391078948975,
|
| 28 |
+
"ssim": 0.5366634726524353,
|
| 29 |
+
"lpips": 0.4511644244194031,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 22.013461538461534,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 8.013461538461538
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_039.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 722,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 39,
|
| 13 |
+
"prompt": "A vintage sewing machine in operation, featuring an antique wooden cabinet and metal parts with a slight patina from age. The needle moves up and down rhythmically as the machine hums softly. A roll of fabric is fed through the machine, creating neat stitches. The background shows an old-fashioned sewing room with wooden floors, a few scattered sewing tools, and a faded floral wallpaper. The focus is on the detailed mechanical workings of the sewing machine. Medium close-up shot, static camera.",
|
| 14 |
+
"original_prompt": "Sewing machine, old sewing machine working.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/039.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2807.0742149353027,
|
| 22 |
+
"excluded_context_kv_latency_ms": 788.0753479003906,
|
| 23 |
+
"wall_generation_s": 5.352534055709839,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.04297858849167824,
|
| 27 |
+
"psnr": 13.667478561401367,
|
| 28 |
+
"ssim": 0.4409210681915283,
|
| 29 |
+
"lpips": 0.47789356112480164,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.36111111111111,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.3611111111111125
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_041.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 724,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 41,
|
| 13 |
+
"prompt": "Close-up macro shot of several large, plump purple plums rotating slowly on a turntable. As the plums spin, water droplets form and glisten on their smooth skins, catching the light and reflecting tiny rainbows. The plums are isolated against a pristine white background, emphasizing their vibrant color and glossy texture. The camera focuses closely, capturing every detail of the fruit's surface as they continue to rotate.",
|
| 14 |
+
"original_prompt": "Few big purple plums rotating on the turntable. water drops appear on the skin during rotation. isolated on the white background. close-up. macro.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/041.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2735.6683959960938,
|
| 22 |
+
"excluded_context_kv_latency_ms": 791.2792129516602,
|
| 23 |
+
"wall_generation_s": 5.28937840461731,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.09740979224443436,
|
| 27 |
+
"psnr": 10.113973617553711,
|
| 28 |
+
"ssim": 0.47013017535209656,
|
| 29 |
+
"lpips": 0.5783026218414307,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.30726495726496,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.307264957264957
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_042.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 725,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 42,
|
| 13 |
+
"prompt": "A stunning young woman with vampire-inspired makeup, featuring pale skin, dark eye shadow, and red lipstick. She has red contact lenses that give her an intense, blood-red gaze. Her hair is flowing freely, styled in soft waves that frame her face. She stands confidently with a slight tilt of her head, exuding an air of mystery and allure. The background is dimly lit with shadows casting an eerie glow. Medium close-up shot focusing on her face and upper body.",
|
| 14 |
+
"original_prompt": "Vampire makeup face of beautiful girl, red contact lenses.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/042.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2745.5383834838867,
|
| 22 |
+
"excluded_context_kv_latency_ms": 786.4932556152344,
|
| 23 |
+
"wall_generation_s": 5.285989999771118,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.02565404586493969,
|
| 27 |
+
"psnr": 15.908441543579102,
|
| 28 |
+
"ssim": 0.605218231678009,
|
| 29 |
+
"lpips": 0.30548080801963806,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.500213675213672,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.500213675213675
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_043.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 726,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 43,
|
| 13 |
+
"prompt": "Close-up shot of a cluttered ashtray filled with cigarette butts on a dimly lit table against a black background. Wisps of smoke curl and flow upwards from the ashtray, creating a hazy atmosphere. The ashtray is old and worn, with visible burn marks and residue. The cigarette butts vary in size and color, some still smoldering slightly. The focus is on the intricate details of the butts and the swirling smoke, with the black background emphasizing the textures and lighting.",
|
| 14 |
+
"original_prompt": "Ashtray full of butts on table, smoke flowing on black background, close-up",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/043.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2760.702377319336,
|
| 22 |
+
"excluded_context_kv_latency_ms": 789.1863632202148,
|
| 23 |
+
"wall_generation_s": 5.311021566390991,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.029001450166106224,
|
| 27 |
+
"psnr": 15.375802516937256,
|
| 28 |
+
"ssim": 0.5335606932640076,
|
| 29 |
+
"lpips": 0.3660205006599426,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.89252136752137,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.892521367521368
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_044.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 727,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 44,
|
| 13 |
+
"prompt": "A serene Pacific coast scene featuring the picturesque town of Carmel-by-the-Sea. The video opens with a wide shot of the vast ocean, showcasing rolling waves crashing against the rocky shoreline. The camera then pans right, revealing the sandy beach dotted with seagulls and a few lone figures walking along the water's edge. In the background, the charming coastal architecture of Carmel-by-the-Sea can be seen, with quaint houses and colorful cottages lining the streets. The sky is a mix of soft pastel hues, reflecting the tranquility of the scene. The overall atmosphere is calm and peaceful, with a gentle breeze blowing over the waves.",
|
| 14 |
+
"original_prompt": "Pacific coast, carmel by the sea ocean and waves.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/044.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2760.8210639953613,
|
| 22 |
+
"excluded_context_kv_latency_ms": 783.3366317749023,
|
| 23 |
+
"wall_generation_s": 5.302945137023926,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.035098686814308167,
|
| 27 |
+
"psnr": 14.547091722488403,
|
| 28 |
+
"ssim": 0.5242984890937805,
|
| 29 |
+
"lpips": 0.38125309348106384,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.343803418803418,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.343803418803419
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_045.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 728,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 45,
|
| 13 |
+
"prompt": "In a vibrant scene set in New York City's Times Square, a large, friendly teddy bear is playfully interacting with a full drum kit. The teddy bear, with its soft brown fur and big, expressive black eyes, sits on a stool, energetically beating the drums with toy-sized drumsticks. Its arms move rhythmically, creating a playful and whimsical atmosphere. The background showcases the bustling energy of Times Square, with bright lights, billboards, and a mix of pedestrians walking past. The scene captures a close-up view, focusing on the teddy bear's animated performance and the colorful surroundings.",
|
| 14 |
+
"original_prompt": "A teddy bear is playing drum kit in NYC Times Square.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/045.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2748.195041656494,
|
| 22 |
+
"excluded_context_kv_latency_ms": 786.1352462768555,
|
| 23 |
+
"wall_generation_s": 5.29471492767334,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.11377016454935074,
|
| 27 |
+
"psnr": 9.439716339111328,
|
| 28 |
+
"ssim": 0.266719251871109,
|
| 29 |
+
"lpips": 0.5168546438217163,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.29273504273504,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.292735042735043
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_047.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 730,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 47,
|
| 13 |
+
"prompt": "A high-definition action scene featuring Tony Stark, aka Iron Man, in his iconic gold and red suit, playing a high-energy electronic guitar. He is standing confidently with a relaxed yet focused expression, swaying slightly to the music. His fingers move swiftly over the guitar strings, creating electrifying sounds. The background showcases a futuristic tech lab with holographic displays and glowing screens. The camera is positioned at a medium close-up angle, capturing the intensity of the performance and the intricate details of his suit.",
|
| 14 |
+
"original_prompt": "An Iron man is playing the electronic guitar, high electronic guitar.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/047.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2761.320182800293,
|
| 22 |
+
"excluded_context_kv_latency_ms": 789.1146545410156,
|
| 23 |
+
"wall_generation_s": 5.308493375778198,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.0880761668086052,
|
| 27 |
+
"psnr": 10.55141568183899,
|
| 28 |
+
"ssim": 0.3219660818576813,
|
| 29 |
+
"lpips": 0.6114063262939453,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.28418803418804,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.284188034188035
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_048.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 731,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 48,
|
| 13 |
+
"prompt": "A playful raccoon is seen playing an electronic guitar, strumming the strings with its front paws. The raccoon has distinctive black facial markings and a bushy tail. It sits comfortably on a small stool, its body slightly tilted as it focuses intently on the instrument. The setting is a cozy, dimly lit room with vintage posters on the walls, adding a retro vibe. The raccoon's expressive eyes convey a sense of joy and concentration. Medium close-up shot, focusing on the raccoon's face and hands interacting with the guitar.",
|
| 14 |
+
"original_prompt": "A raccoon is playing the electronic guitar.",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/048.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2793.2196159362793,
|
| 22 |
+
"excluded_context_kv_latency_ms": 786.4256286621094,
|
| 23 |
+
"wall_generation_s": 5.338623285293579,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.056047581136226654,
|
| 27 |
+
"psnr": 12.514431476593018,
|
| 28 |
+
"ssim": 0.39509251713752747,
|
| 29 |
+
"lpips": 0.5222867131233215,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.696153846153845,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.696153846153847
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_050.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 733,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 50,
|
| 13 |
+
"prompt": "An artistic depiction of a corgi's head as an explosion of colorful nebula gases. The corgi\u2019s head features distinct facial elements such as large, expressive eyes, floppy ears, and a friendly smile, all rendered in swirling cosmic colors. The nebula includes vibrant hues of blues, purples, and pinks, creating a fantastical and whimsical scene. The background fades into a starry night sky, adding depth and scale to the composition. The corgi\u2019s head occupies the center of the frame, with the nebula expanding outward in a symmetrical explosion. The overall style is imaginative and detailed, capturing the playful essence of a corgi through a celestial lens. Full-frame close-up view.",
|
| 14 |
+
"original_prompt": "A corgi's head depicted as an explosion of a nebula",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/050.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2722.8430099487305,
|
| 22 |
+
"excluded_context_kv_latency_ms": 786.681022644043,
|
| 23 |
+
"wall_generation_s": 5.263099908828735,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.06016024202108383,
|
| 27 |
+
"psnr": 12.20690369606018,
|
| 28 |
+
"ssim": 0.39258873462677,
|
| 29 |
+
"lpips": 0.390608012676239,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.42457264957265,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.424572649572651
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_052.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 735,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 52,
|
| 13 |
+
"prompt": "A futuristic cityscape where humans have achieved teleportation technology. In this scene, several diverse individuals from various races and ages stand in a bustling teleportation hub. They wear sleek, modern clothing and futuristic accessories. The teleportation devices emit soft, glowing blue lights as people step into them, disappearing in a burst of light. The background showcases towering skyscrapers, flying vehicles, and holographic advertisements, creating a vibrant and advanced urban environment. Wide shot, static camera to capture the dynamic activity and technological marvels.",
|
| 14 |
+
"original_prompt": "A future where humans have achieved teleportation technology",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/052.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2695.1482696533203,
|
| 22 |
+
"excluded_context_kv_latency_ms": 783.1365814208984,
|
| 23 |
+
"wall_generation_s": 5.23565411567688,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.09916526824235916,
|
| 27 |
+
"psnr": 10.036404132843018,
|
| 28 |
+
"ssim": 0.30401748418807983,
|
| 29 |
+
"lpips": 0.5851956605911255,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.497649572649575,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.497649572649574
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_053.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 736,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 53,
|
| 13 |
+
"prompt": "A serene underwater scene featuring a translucent, bioluminescent jellyfish gently floating through the vast ocean. The jellyfish has soft, undulating tentacles that emit a gentle glow, creating mesmerizing patterns as they sway gracefully in the water. The surrounding ocean is filled with ambient light filtering through the surface, casting a tranquil blue hue over the scene. Various small fish and plankton drift by in the background, adding to the sense of life and movement in the deep sea. The camera maintains a medium close-up perspective, focusing on the jellyfish's ethereal presence.",
|
| 14 |
+
"original_prompt": "A jellyfish floating through the ocean, with bioluminescent tentacles",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/053.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2751.742572784424,
|
| 22 |
+
"excluded_context_kv_latency_ms": 787.0065536499023,
|
| 23 |
+
"wall_generation_s": 5.2983903884887695,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.008874152787029743,
|
| 27 |
+
"psnr": 20.51873207092285,
|
| 28 |
+
"ssim": 0.7336246967315674,
|
| 29 |
+
"lpips": 0.3145241141319275,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.308760683760685,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.308760683760686
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_054.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 737,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 54,
|
| 13 |
+
"prompt": "A realistic animation of a Mars rover navigating across the rugged Martian terrain. The rover is equipped with solar panels, antennas, and various scientific instruments. It moves slowly and steadily over large boulders and fine red dust, leaving tire tracks behind. The landscape is dotted with craters, rocky outcrops, and barren plains under a pale orange sky. The camera follows the rover from a mid-shot perspective as it continues its exploration mission. Static scene, focusing on the rover's movement and interaction with the environment.",
|
| 14 |
+
"original_prompt": "A Mars rover moving on Mars",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/054.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2736.3347663879395,
|
| 22 |
+
"excluded_context_kv_latency_ms": 785.4596557617188,
|
| 23 |
+
"wall_generation_s": 5.280845403671265,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.026303844526410103,
|
| 27 |
+
"psnr": 15.79980731010437,
|
| 28 |
+
"ssim": 0.6153629422187805,
|
| 29 |
+
"lpips": 0.39109376072883606,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.6198717948718,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.619871794871795
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_058.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 741,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 58,
|
| 13 |
+
"prompt": "A futuristic, towering giant robot standing in the neon-lit streets of Cyberpunk Beijing. The robot is sleek and metallic, with intricate cybernetic designs and glowing LED lights running along its body. It has a powerful, muscular build and stands atop a bustling street filled with holographic advertisements and futuristic vehicles. The cityscape behind it includes towering skyscrapers with illuminated windows and billowing smokestacks, creating a vibrant and dynamic urban environment. The robot's eyes are highlighted and emit a soft blue glow, adding to its imposing presence. Wide shot capturing the robot and the bustling cityscape below.",
|
| 14 |
+
"original_prompt": "A super cool giant robot in Cyberpunk Beijing",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/058.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2693.3704109191895,
|
| 22 |
+
"excluded_context_kv_latency_ms": 784.286262512207,
|
| 23 |
+
"wall_generation_s": 5.231689691543579,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.06754936277866364,
|
| 27 |
+
"psnr": 11.703786849975586,
|
| 28 |
+
"ssim": 0.29188215732574463,
|
| 29 |
+
"lpips": 0.547601044178009,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.35683760683761,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.356837606837608
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_059.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 742,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 59,
|
| 13 |
+
"prompt": "A serene tropical beach at sunrise, where the first rays of sunlight illuminate the golden sand and crystal-clear turquoise water. In the foreground, tall palm trees sway gently in the breeze, casting long shadows across the sand. The water is calm and inviting, reflecting the vibrant hues of the early morning sky. The scene is peaceful and tranquil, with soft waves lapping at the shore. Wide-angle shot, capturing the expansive beauty of the beach and the horizon where the sun is just beginning to rise.",
|
| 14 |
+
"original_prompt": "A tropical beach at sunrise, with palm trees and crystal-clear water in the foreground",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/059.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2764.791877746582,
|
| 22 |
+
"excluded_context_kv_latency_ms": 789.6010513305664,
|
| 23 |
+
"wall_generation_s": 5.315661191940308,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.0364266075193882,
|
| 27 |
+
"psnr": 14.385812282562256,
|
| 28 |
+
"ssim": 0.44484084844589233,
|
| 29 |
+
"lpips": 0.44442710280418396,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.513247863247862,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.513247863247863
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_061.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 744,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 61,
|
| 13 |
+
"prompt": "A young woman named Gwen Stacy, with shoulder-length curly auburn hair and wearing glasses, sits comfortably in a cozy armchair, engrossed in reading a book. She has fair skin and wears a casual outfit consisting of a pastel-colored blouse and jeans. Her expression is serene and focused, with a hint of concentration as she turns the pages gently. The room around her is softly lit with warm hues, featuring a wooden desk with scattered books and a small lamp. The background includes a bookshelf filled with various books and a window letting in natural light. Medium close-up shot, static scene.",
|
| 14 |
+
"original_prompt": "Gwen Stacy reading a book",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/061.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2794.995990753174,
|
| 22 |
+
"excluded_context_kv_latency_ms": 787.2070465087891,
|
| 23 |
+
"wall_generation_s": 5.343601226806641,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.0317816324532032,
|
| 27 |
+
"psnr": 14.978238344192505,
|
| 28 |
+
"ssim": 0.47715815901756287,
|
| 29 |
+
"lpips": 0.4465090036392212,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.386111111111113,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.386111111111112
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_062.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 745,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 62,
|
| 13 |
+
"prompt": "A high-definition, cinematic scene featuring Tony Stark as Iron Man soaring through the sky. He is clad in his iconic gold and red armored suit, which gleams under the sunlight. His arms are extended slightly forward as he flies, showcasing the intricate design and glowing arc reactor on his chest. The sky is a vivid mix of blue and orange hues, suggesting a sunset. In the background, there are distant skyscrapers and clouds, emphasizing the urban setting. Iron Man's face is partially obscured by his helmet visor, but his determined expression can be seen. The scene is captured in a medium shot, maintaining a steady camera angle as Iron Man continues to fly gracefully.",
|
| 14 |
+
"original_prompt": "Iron Man flying in the sky",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/062.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2736.42728805542,
|
| 22 |
+
"excluded_context_kv_latency_ms": 785.7180404663086,
|
| 23 |
+
"wall_generation_s": 5.282249450683594,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.06932883709669113,
|
| 27 |
+
"psnr": 11.590861082077026,
|
| 28 |
+
"ssim": 0.5518754720687866,
|
| 29 |
+
"lpips": 0.48677363991737366,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.4732905982906,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.473290598290598
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_063.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 746,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 63,
|
| 13 |
+
"prompt": "Oil painting style, a panoramic view of The Bund in Shanghai at sunset. The scene features iconic buildings such as the Peace Hotel and the Customs House, with their distinctive architectural details illuminated by warm golden hues. The Huangpu River flows gracefully in the foreground, reflecting the vibrant colors of the sky. Crowds of people in modern attire stroll along the riverbank, adding life to the bustling atmosphere. In the background, skyscrapers loom over the historic district, creating a striking contrast between old and new Shanghai. Wide shot, static scene capturing the beauty of the cityscape under a serene twilight sky.",
|
| 14 |
+
"original_prompt": "The bund Shanghai, oil painting",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/063.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2787.317169189453,
|
| 22 |
+
"excluded_context_kv_latency_ms": 788.6123504638672,
|
| 23 |
+
"wall_generation_s": 5.3345348834991455,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.025974079966545105,
|
| 27 |
+
"psnr": 15.8545982837677,
|
| 28 |
+
"ssim": 0.5537792444229126,
|
| 29 |
+
"lpips": 0.3638566732406616,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.56666666666667,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.566666666666667
|
| 37 |
+
}
|
| 38 |
+
}
|
eval_out/per_prompt/sf_motioncache_x1.3/overall_consistency_064.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"status": "complete",
|
| 3 |
+
"protocol": "Self-Forcing Extended-251 Full Evaluation",
|
| 4 |
+
"strategy": "sf_motioncache_x1.3",
|
| 5 |
+
"base_model": "self_forcing",
|
| 6 |
+
"method": "motioncache",
|
| 7 |
+
"param": "thresh",
|
| 8 |
+
"param_value": 0.7559814453125,
|
| 9 |
+
"target_speedup": 1.3,
|
| 10 |
+
"global_index": 747,
|
| 11 |
+
"prompt_suite": "overall_consistency",
|
| 12 |
+
"suite_index": 64,
|
| 13 |
+
"prompt": "A CGI animation of Yoda, the iconic green Jedi master, playing a guitar on a stage. Yoda is dressed in his traditional Jedi robes and has a playful, focused expression as he strums the guitar strings. The stage is dimly lit with spotlights shining down on him, creating dramatic shadows. The background features a live audience with various alien species cheering him on. Yoda is sitting on a stool, leaning slightly forward, and his fingers move nimbly over the guitar. Medium close-up shot focusing on Yoda's face and hands interacting with the guitar.",
|
| 14 |
+
"original_prompt": "Yoda playing guitar on the stage",
|
| 15 |
+
"seed": 0,
|
| 16 |
+
"video": "/local/zoubin/cz/projects/comparison/eval_out/generated_videos/sf_motioncache_x1.3/overall_consistency/064.mp4",
|
| 17 |
+
"num_frames": 81,
|
| 18 |
+
"height": 480,
|
| 19 |
+
"width": 832,
|
| 20 |
+
"fps": 16,
|
| 21 |
+
"policy_latency_ms": 2737.632148742676,
|
| 22 |
+
"excluded_context_kv_latency_ms": 787.0390243530273,
|
| 23 |
+
"wall_generation_s": 5.2834954261779785,
|
| 24 |
+
"reference_strategy": "sf_ffff",
|
| 25 |
+
"pixel_metrics_vs_ffff": {
|
| 26 |
+
"mean_mse": 0.04936559498310089,
|
| 27 |
+
"psnr": 13.065756559371948,
|
| 28 |
+
"ssim": 0.43760061264038086,
|
| 29 |
+
"lpips": 0.4590982496738434,
|
| 30 |
+
"num_frames": 81
|
| 31 |
+
},
|
| 32 |
+
"cache_diagnostics": {
|
| 33 |
+
"denoise_forwards": 28,
|
| 34 |
+
"compute_equivalent_forwards": 21.635897435897437,
|
| 35 |
+
"middle_steps": 14,
|
| 36 |
+
"middle_compute_equivalent": 7.635897435897437
|
| 37 |
+
}
|
| 38 |
+
}
|