HKUST-GZ Open-Sources Motion Forcing for Physics-Aware Driving Video Generation

1 reports archived1 independent sourcesupdated 11/10/2026, 10:22:00

Synthesis & Latest Updates

1 Sources Cross-Validated

A team at HKUST (Guangzhou) introduces Motion Forcing, a framework that decomposes video generation into three layers: points (trajectories), shape (depth), and appearance (color), deciding motion before rendering visuals to eliminate vehicle interpenetration and physical implausibility. On real Waymo driving data it ranks first on both motion coherence and physical plausibility, surpassing Seedance 2.0 and Wan 2.6, with code and an interactive demo open-sourced.

LATEST/A team at HKUST (Guangzhou) introduces Motion Forcing, a framework that decomposes video generation into three layers: points (trajectories), shape (depth), and appearance (color), deciding motion before rendering visuals to eliminate vehicle interpenetration and physical implausibility. On real Waymo driving data it ranks first on both motion coherence and physical plausibility, surpassing Seedance 2.0 and Wan 2.6, with code and an interactive demo open-sourced.

Hourly heat curve

13 fully observed hours

Now

9.0

Peak

11.811:00

24h change

–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

Coverage timeline

Total 1 reports · Latest first

量子位 (微信公众号)T2Nov 10, 10:22·78 pts
HKUST-GZ Open-Sources Motion Forcing for Physics-Aware Driving Video Generation
- Splits generation into points, shape, and appearance layers, fixing motion before rendering visuals to reduce interpenetration and vanishing objects
- Ranks first on motion coherence and physical plausibility on real Waymo data, ahead of Seedance 2.0 and Wan 2.6
- Generates long-tail hazard scenarios such as cut-ins and avoidance from a single image plus a few trajectories; code and demo are open-sourced
View Related Reports & Analysis Original Source