Physics-Guided Motion Loss for Video Generation Model
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866908558850785280 |
|---|---|
| author | Xue, Bowen Guarnera, Giuseppe Claudio Zhao, Shuang Montazeri, Zahra |
| author_facet | Xue, Bowen Guarnera, Giuseppe Claudio Zhao, Shuang Montazeri, Zahra |
| contents | Current video diffusion models generate visually compelling content but often violate basic laws of physics, producing subtle artifacts like rubber-sheet deformations and inconsistent object motion. We introduce a frequency-domain physics prior that improves motion plausibility without modifying model architectures. Our method decomposes common rigid motions (translation, rotation, scaling) into lightweight spectral losses, requiring only 2.7% of frequency coefficients while preserving 97%+ of spectral energy. Applied to Open-Sora, MVDIT, and Hunyuan, our approach improves both motion accuracy and action recognition by ~11% on average on OpenVID-1M (relative), while maintaining visual quality. User studies show 74--83% preference for our physics-enhanced videos. It also reduces warping error by 22--37% (depending on the backbone) and improves temporal consistency scores. These results indicate that simple, global spectral cues are an effective drop-in regularizer for physically plausible motion in video diffusion. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_02244 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Physics-Guided Motion Loss for Video Generation Model Xue, Bowen Guarnera, Giuseppe Claudio Zhao, Shuang Montazeri, Zahra Computer Vision and Pattern Recognition Artificial Intelligence Current video diffusion models generate visually compelling content but often violate basic laws of physics, producing subtle artifacts like rubber-sheet deformations and inconsistent object motion. We introduce a frequency-domain physics prior that improves motion plausibility without modifying model architectures. Our method decomposes common rigid motions (translation, rotation, scaling) into lightweight spectral losses, requiring only 2.7% of frequency coefficients while preserving 97%+ of spectral energy. Applied to Open-Sora, MVDIT, and Hunyuan, our approach improves both motion accuracy and action recognition by ~11% on average on OpenVID-1M (relative), while maintaining visual quality. User studies show 74--83% preference for our physics-enhanced videos. It also reduces warping error by 22--37% (depending on the backbone) and improves temporal consistency scores. These results indicate that simple, global spectral cues are an effective drop-in regularizer for physically plausible motion in video diffusion. |
| title | Physics-Guided Motion Loss for Video Generation Model |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2506.02244 |