Video World Models · 2026-10-30

FutureWorld: A Plug-and-Play Causal Next-State Predictor for Interactive Video World Exploration

Guangyuan Li, Kelu Yao, Shuolin Xu, Xinlei Yu, Wei Xing, Lei Zhao, Peng-Tao Jiang, Xiaogang Xu, Xiaobin Hu

arXiv:2610.39619PDF

AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.

TL;DR

FutureWorld introduces a plug-and-play Causal Next-Chunk Predictor (C-NCP) that predicts the next latent state from the current generated latent and the next action, then modulates a frozen autoregressive video DiT to generate the next chunk. The authors aim to reduce error recirculation from reusing imperfect generated history, improving long-horizon interactive consistency across multiple generators and benchmarks.

Why it matters

Interactive video world models must follow actions while keeping scene structure and semantics over long interactions. The authors identify “error recirculation” from history reuse and propose an explicit future-state prior that guides generation before the next chunk becomes history.

Method

  • Learn a future-state prior from real video transitions to predict the complete next latent chunk from current latent and next action (C-NCP).
  • Use VAE-specific input adapters plus a shared causal spatiotemporal predictor for cross-model adaptability; adapt predictions back to each model’s latent space.
  • Apply prediction-guided, shape-preserving channel-wise residual modulation (residual feature modulation with adaptive instance normalization-style scale/shift) to guide a frozen autoregressive video DiT during inference.
Abstract (from arXiv)

Interactive video world models aim to follow user actions while maintaining scene consistency over extended interactions. History-based methods reuse generated latents to preserve scene content, but three limitations of this reuse contribute to error recirculation. 1) The latent generated at each interaction is retained as history, potentially containing blurred details, geometric distortions, or semantic drift. 2) Although retrieved historical content matches the current scene, it can still contain errors that are carried into the next chunk. 3) New outputs then become history for later interactions, potentially reinforcing the same errors through repeated reuse. To limit this cycle, we aim to predict the future scene state from the current generated latent and the next action before generating the next interaction's video chunk. This future-state prior provides guidance for the next chunk, with the aim of reducing the errors carried forward from history. To this end, we propose a plug-and-play Causal Next-Chunk Predictor (C-NCP) with cross-model adaptability and prediction-guided modulation. We integrate C-NCP into four base generators and evaluate the augmented models across three benchmarks for interactive long-video generation and action-conditioned world modeling. Extensive experiments demonstrate reduced error accumulation and improved long-horizon consistency across diverse generation settings. The project page is at https://GuangYuanKK.github.io/futureworld.

Related papers

Publication record

Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.

First announced by arXiv: 2026-10-30 · Indexed by Pub-AI: 2026-10-30 UTC

preprint

FutureWorld: A Plug-and-Play Causal Next-State Predictor for Interactive Video World Exploration

Guangyuan Li, Kelu Yao, Shuolin Xu, Xinlei Yu, Wei Xing, Lei Zhao, Peng-Tao Jiang, Xiaogang Xu, Xiaobin Hu

First announced by arXiv: 2026-10-30 · Original record ↗

arxiv
2610.39619
Explicit source versions (1)
  • v1 · Date not supplied
  1. Preprint available

    Source date: 2026-10-30 · First detected: 2026-10-30 UTC · Historical baseline

    Source evidence (1)
    • Official evidence ↗ · RSS guid=oai:arXiv.org:2610.39619v1; arxiv:announce_type=new; item pubDate (official metadata)
Abstract · arxiv

Interactive video world models aim to follow user actions while maintaining scene consistency over extended interactions. History-based methods reuse generated latents to preserve scene content, but three limitations of this reuse contribute to error recirculation. 1) The latent generated at each interaction is retained as history, potentially containing blurred details, geometric distortions, or semantic drift. 2) Although retrieved historical content matches the current scene, it can still contain errors that are carried into the next chunk. 3) New outputs then become history for later interactions, potentially reinforcing the same errors through repeated reuse. To limit this cycle, we aim to predict the future scene state from the current generated latent and the next action before generating the next interaction's video chunk. This future-state prior provides guidance for the next chunk, with the aim of reducing the errors carried forward from history. To this end, we propose a plug-and-play Causal Next-Chunk Predictor (C-NCP) with cross-model adaptability and prediction-guided modulation. We integrate C-NCP into four base generators and evaluate the augmented models across three benchmarks for interactive long-video generation and action-conditioned world modeling. Extensive experiments demonstrate reduced error accumulation and improved long-horizon consistency across diverse generation settings. The project page is at https://GuangYuanKK.github.io/futureworld.