Overview

Video world models predict how scenes change over time, often in response to actions or camera movements, so agents can explore, plan, and test decisions in imagined environments. Common approaches use autoregressive or diffusion video generation, sometimes predicting compact visual features instead of pixels, and add action conditioning, 3D geometry, persistent memory, or explicit world states to improve control and consistency. Applications include robotic manipulation, autonomous driving, navigation, and interactive games, with evaluation increasingly testing physical dynamics, action following, memory, and downstream task performance alongside visual quality. A central challenge is that realistic-looking video does not necessarily preserve hidden state, obey physics, or support reliable long-horizon interaction, while richer simulation and memory can raise compute and latency costs.

Latest papers

CardsList

285 papers37 papers in the last four weeks, down 40% on the four weeks before. 0.3% of all new papers.

  1. ShareVerse: Multi-Agent Consistent Video Generation for Shared World Modeling

    Mar 3, 2026Jiayi Zhu, Jianing Zhang, Yiying Yang +2Video World ModelsMulti-View Video Generation

  2. DreamWorld: Unified World Modeling in Video Generation

    Feb 28, 2026Boming Tan, Xiangdong Zhang, Ning Liao +73D-Aware Video GenerationVideo World Models

  3. UCM: Unified Modeling of Camera Control and Memory with Time-aware Positional Encoding Warping for World Models

    Feb 26, 2026Tianxing Xu, Zixuan Wang, Guangyuan Wang +5Camera-Controlled Video GenerationVideo Diffusion Models

  4. RAYNOVA: Scale-Temporal Autoregressive World Modeling in Ray Space

    Feb 24, 2026Yichen Xie, Chensheng Peng, Mazen Abdelfattah +6Multi-View Video GenerationWorld Models

  5. When World Models Dream Wrong: Physical-Conditioned Adversarial Attacks against World Models

    Feb 21, 2026Zhixiang Guo, Siyuan Liang, Andras Balogh +4Adversarial Attacks on World ModelsAdversarial Attacks

  6. WoVR: World Models as Reliable Simulators for Post-Training VLA Policies with RL

    Feb 15, 2026Zhennan Jiang, Shangqing Zhou, Yutong Jiang +11World Models for RoboticsWorld Models

  7. Flow Equivariant World Models: Structured Memory for Dynamic Environments

    Jan 3, 2026Hansen Jin Lillemark, Benhao Huang, Fangneng Zhan +2Memory-Augmented World ModelsWorld Model Learning

  8. CustomX: Unified Character, Action, and Scene Customization in Video World Models

    Dec 18, 2025Yitong Wang, Fangyun Wei, Hongyang Zhang +2HOI GenerationAction-Conditioned Video Generation

  9. WorldWander: Bridging Egocentric and Exocentric Worlds in Video Generation

    Nov 27, 2025Quanjian Song, Yiren Song, Kelly Peng +2Video Diffusion ModelsVideo World Models