ICML oral paper on motion attribution for video generation released

The ICML 2026 oral talk on Motion Attribution for Video Generation is now online, introducing Motive, a framework that traces how training samples shape motion in generated videos to probe the model's internals, winning 74.1% in human evaluation.

2026-09-04 ~ 2026-09-04 · 2 related posts

• ICML oral Motive: first motion attribution framework for video generation, 74.1% win rate — cindy_x_wu · 2026-09-04
• ICML 2026 oral talk live: motion attribution for video generation — cindy_x_wu · 2026-09-04