WorldPlay: Towards Long-Term Geometric Consistency for Real-Time Interactive World Modeling
By: Wenqiang Sun, Haiyu Zhang, Haoyuan Wang, Junta Wu, Zehan Wang, Zhenwei Wang, Yunhong Wang, Jun Zhang, Tengfei Wang, Chunchao Guo
Published: 2025-12-17
View on arXiv →#cs.AI
Abstract
This paper presents WorldPlay, a streaming video diffusion model that enables real-time, interactive world modeling with long-term geometric consistency. It resolves the trade-off between speed and memory through innovations like Dual Action Representation, Reconstituted Context Memory, and Context Forcing, generating long-horizon streaming 720p video at 24 FPS with superior consistency and strong generalization across diverse scenes.