Motion Attribution for Video Generation

TL;DR

Motive is a novel gradient-based framework that attributes motion influence in video generation models, enabling better data curation and improved temporal dynamics in generated videos.

Contribution

This paper introduces the first motion-specific data attribution method for video generation, facilitating targeted data curation to enhance motion quality.

Findings

Identifies clips that significantly affect motion in video models.

Improves motion smoothness and dynamic range in generated videos.

Achieves 74.1% human preference over baseline models.

Abstract

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.