Original Tweet
New NVIDIA Paper
We introduce Motive, a motion-centric, gradient-based data attribution method that traces which training videos help or hurt video generation.
By isolating temporal dynamics from static appearance, Motive identifies which training videos shape motion in video generation.
๐ https://research.nvidia.com/labs/sil/projects/MOTIVE/
1/10
๐ ์๋ณธ ๋งํฌ
๋ฏธ๋์ด
๐ Related
- 360anything-geometry-free-lifting-of-images-and-videos-to-360 โ ์ฃผ์ : World Model
- do-we-really-need-an-external-world-model โ ์ฃผ์ : World Model
- exciting-new-work-on-detailed-pixel-level-dense-3d-visual-understanding-of-video โ ์ฃผ์ : World Model
- its-been-a-busy-week-for-world-models-and-their-apis-httpstcondhlgquv6o โ ์ฃผ์ : World Model
- we-introduce-egowm-a-video-world-model-that-simulates-eve-1x-humanoid-interactio โ ์ฃผ์ : World Model