Jun 2026
Directing the World: Fast Autoregressive Video Generation with Compositional Human-Camera Control
This work proposes "Directing the World", a fast autoregressive framework for controllable world-model video generation with compositional human-motion and camera-trajectory control, and introduces a Fast-Slow Memory training strategy to stabilize long-horizon rollout learning and improve convergence.
Haoyuan Wang, Yabo Chen, Haibin Huang et al.
· arXiv.org · 1 citation