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. 🔗 1/10

Motion is the defining element of videos. Unlike image generation, which produces a single frame, video generation needs to capture how objects move, interact, and obey physical constraints. Understanding the source of motion helps us interpret, diagnose, and curate data for better video generation models. 2/10

Our key idea: compute gradients through the diffusion model, and weight them by motion magnitude. This emphasizes moving regions (such as trajectories or deformations) and downplays static backgrounds. 3/10

We extract motion masks using AllTracker and compute motion-weighted influence scores. The result is a ranking of which training clips most strongly affect a model’s temporal behavior. 4/10

Using these influence scores, we select only the most motion-relevant data for fine-tuning. With only 10% of the data, we surpass full-dataset results. Motive improves dynamic degree and motion smoothness on VBench while preserving visual and semantic quality. 5/10

In human studies, videos generated with Motive-selected data were preferred 74.1% of the time over the base model and 53.1% over full fine-tuning. Participants found the motion smoother and more physically consistent. 6/10

Motive reveals that video models share influential training data across related motion types. High-overlap pairs like bounce-float suggest shared underlying dynamics, while low-overlap pairs like free fall and stretch indicate distinct learning preferences. These patterns hold consistently across datasets, suggesting certain motion types share transferable training signals. 7/10

Motive suggests a new axis for scaling video generation data: not how much data we use, but which motion patterns we train on. This motion-level attribution helps us understand how video diffusion models internalize motion dynamics. If we can trace which experiences teach which dynamics, we move closer to treating training data as a first-class citizen, and this raises interesting questions for data mixture and coverage. This perspective could inform how we build, scale, and curate data for next-gen video world models. 8/10

Motive is inspired by a growing body of work on motion, world modeling, and data attribution, from large-scale video diffusion (Wan, @Alibaba_Wan, Veo, Cosmos), to influence-based analysis, such as TRAK ( @kris_georgiev1, @andrew_ilyas), systematic video evaluation efforts such as VBench ( @liuziwei7), dense point tracking (AllTracker, @AdamWHarley), and many others. We’re excited to build on these ideas and push data-centric tools toward understanding how motion is learned. 9/10

Huge thanks to my amazing coauthors @paschalidoud_1 @JunGao33210520 Antonio Torralba @lealtaixe @orussakovsky @FidlerSanja @jonLorraine9, I learned so much working with them! 💗 Check out more cool research at: @NVIDIAAI @PrincetonCS @MIT_CSAIL @UofTCompSci @VectorInst 10/10

Congrats Xindi!!

Thank you Amir!!!

Cool paper, would be interested in figuring out what the analogy of this would be for language.

This was a great read. Thanks Xindi

Congrats — fantastic work on data attribution!

Cool work congrats!

Thank you Zeeshan!!

Congrats!

Thanks Ethan!

Wonderful work! Congratulations.

Thank you!! 😊

Motive 提出一種基於梯度的動態數據歸因方法，能有效區分訓練影片中時間動態與靜態特徵，進而精準識別對動態生成有正向或負向影響的訓練內容，為視頻生成的可解釋性提供新維度