Motion Attribution for Video Generation
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866917201725882368 |
|---|---|
| author | Wu, Xindi Paschalidou, Despoina Gao, Jun Torralba, Antonio Leal-Taixé, Laura Russakovsky, Olga Fidler, Sanja Lorraine, Jonathan |
| author_facet | Wu, Xindi Paschalidou, Despoina Gao, Jun Torralba, Antonio Leal-Taixé, Laura Russakovsky, Olga Fidler, Sanja Lorraine, Jonathan |
| contents | Despite the rapid progress of video generation models, the role of data in influencing motion is poorly understood. We present Motive (MOTIon attribution for Video gEneration), a motion-centric, gradient-based data attribution framework that scales to modern, large, high-quality video datasets and models. We use this to study which fine-tuning clips improve or degrade temporal dynamics. Motive isolates temporal dynamics from static appearance via motion-weighted loss masks, yielding efficient and scalable motion-specific influence computation. On text-to-video models, Motive identifies clips that strongly affect motion and guides data curation that improves temporal consistency and physical plausibility. With Motive-selected high-influence data, our method improves both motion smoothness and dynamic degree on VBench, achieving a 74.1% human preference win rate compared with the pretrained base model. To our knowledge, this is the first framework to attribute motion rather than visual appearance in video generative models and to use it to curate fine-tuning data. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_08828 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Motion Attribution for Video Generation Wu, Xindi Paschalidou, Despoina Gao, Jun Torralba, Antonio Leal-Taixé, Laura Russakovsky, Olga Fidler, Sanja Lorraine, Jonathan Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning Multimedia Robotics 68T45 I.2.10; I.4.8; I.2.6; I.5.1 Despite the rapid progress of video generation models, the role of data in influencing motion is poorly understood. We present Motive (MOTIon attribution for Video gEneration), a motion-centric, gradient-based data attribution framework that scales to modern, large, high-quality video datasets and models. We use this to study which fine-tuning clips improve or degrade temporal dynamics. Motive isolates temporal dynamics from static appearance via motion-weighted loss masks, yielding efficient and scalable motion-specific influence computation. On text-to-video models, Motive identifies clips that strongly affect motion and guides data curation that improves temporal consistency and physical plausibility. With Motive-selected high-influence data, our method improves both motion smoothness and dynamic degree on VBench, achieving a 74.1% human preference win rate compared with the pretrained base model. To our knowledge, this is the first framework to attribute motion rather than visual appearance in video generative models and to use it to curate fine-tuning data. |
| title | Motion Attribution for Video Generation |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning Multimedia Robotics 68T45 I.2.10; I.4.8; I.2.6; I.5.1 |
| url | https://arxiv.org/abs/2601.08828 |