ConsistWorld enables consistent multi-agent video streams from one image
ConsistWorld: Evidence Routing for Consistent Multi-Agent World Models
Computer Vision and Pattern Recognition
Summary
Generating videos that show a scene from multiple viewpoints controlled by different agents is challenging because each agent only sees part of the story and at different times. The authors present ConsistWorld, which creates video streams from a single shared image and uses clever methods to make views from different agents agree with each other over time. It does this by smartly choosing past information and deciding what peers currently see to keep the video consistent and coherent. This helps create long videos that look coordinated across viewers while still being high quality.
What this means in practice
- •For virtual reality developers: Create consistent multi-view VR experiences where multiple users see a static scene from their own synchronized viewpoints without temporal glitches.
- •For surveillance system engineers: Generate coordinated video streams from multiple cameras monitoring a static environment, ensuring temporal and spatial consistency across views.
Authors
Qianxun Xu, Xianfang Zeng, Xinyao Liao, Wei Cheng, Gang Yu, Chi Zhang
Abstract
Autoregressive video world models enable temporally coherent generation for a single observer. Extending them to multiple agents requires consistency across independently controlled views and temporal gaps under causal streaming. We present ConsistWorld, a multi-agent world model that generates camera-controlled video streams of a static scene from one shared image. We formulate consistency as routing evidence from committed multi-agent history and concurrently generated peer views to the tokens being generated. Pose Conditioned Memory Retrieval selects relevant historical observations from all agents, recovering evidence beyond the recent context window. Visibility-Gated Peer Sharing regulates current peer information according to estimated historical coverage and current-view overlap. Together, they determine which historical observations enter the context and where concurrent peer information contributes, supporting long-term recall and coordinated exploration. Both mechanisms use camera geometry and maintain a bounded active context for a fixed agent count and retrieval budget. Experiments on evidence sharing cases and video length and agent number generalizations show that ConsistWorld achieves a strong cross-time and cross-agent consistency while preserving competitive generation quality.