Agentic Physics-Grounded Video Generation via Memory-Augmented In-Context Reasoning

📄 arXiv: 2610.23677v1 📥 PDF

作者: Xikang Guan, Shaocong Yang, Yanqi Bao, Can Wang, Zheng Gu, Zhengqing Chen, Kai Chen, Jing Huo, Yang Gao

分类: cs.CV

发布日期: 2026-10-16

备注: 12 pages, 5 figures, 1 table

🔗 代码/项目: PROJECT_PAGE


💡 一句话要点

提出PhysRepair框架以解决视频生成中的物理不一致性问题

🎯 匹配领域: 支柱四:生成式动作 (Generative Motion) 支柱八:物理动画 (Physics-based Animation)

关键词: 视频生成 物理合理性 因果推理 经验增强学习 局部修复 动态多样性 生成-诊断-修复

📋 核心要点

  1. 现有视频生成方法在物理指导上存在局限,无法有效处理生成后出现的物理不一致性。
  2. 本文提出PhysRepair框架,通过因果推理与经验增强学习,实现对物理不一致性的局部修复。
  3. 实验结果显示,PhysRepair在固体力学、流体动力学等多个领域的物理合理性上优于现有方法。

📝 摘要(中文)

物理合理性对真实视频生成至关重要。现有方法在合成前通过模拟或精细提示施加物理指导,但这限制了动态多样性,可能遗漏生成后才显现的错误,并在不一致时需要重复生成。为此,本文研究了一种生成-诊断-修复的范式,提出了PhysRepair框架,结合因果推理与修复经验,实现局部自适应修复。实验结果表明,PhysRepair在多个物理领域上显著提升了生成视频的物理合理性。

🔬 方法详解

问题定义:本文旨在解决视频生成中物理不一致性的问题。现有方法在合成前施加物理指导,导致动态多样性受限,且无法处理生成后出现的错误。

核心思路:PhysRepair框架通过生成-诊断-修复的过程,结合因果推理与修复经验,能够在生成后进行局部修复,保持原始生成能力。

技术框架:该框架主要包括两个模块:1) 时空因果约束,用于识别无效物理状态及其因果影响状态;2) 经验增强的上下文学习,将修复上下文与以往修复经验结合,适应不同的物理失败。

关键创新:PhysRepair的创新在于其生成-诊断-修复的范式,能够在生成后进行选择性修复,显著提升物理合理性,与传统方法的预先指导形成鲜明对比。

关键设计:在技术细节上,框架中采用了时空因果约束来指导局部再生,同时利用经验增强学习来积累修复经验,优化修复决策。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

实验结果表明,PhysRepair在Physics-IQ验证集上显著提升了生成视频的物理合理性,相较于现有方法,物理合理性提升幅度达到20%以上,尤其在固体力学和流体动力学领域表现突出。

🎯 应用场景

该研究的潜在应用领域包括虚拟现实、动画制作和游戏开发等,能够为生成更加真实的物理场景提供技术支持。未来,PhysRepair框架有望在更广泛的多模态生成任务中发挥作用,提升生成内容的物理合理性和视觉真实感。

📄 摘要(原文)

Physical plausibility is essential for realistic video generation. Existing methods impose physical guidance before synthesis via simulations or refined prompts, but this can limit dynamic diversity, miss errors that emerge only after generation, and require repeated regeneration when inconsistencies remain. We therefore study a generate-diagnose-repair paradigm that preserves the original generation capability and selectively repairs physical inconsistencies afterward, raising two key questions: what to repair and how to repair? To address them, we introduce PhysRepair, an agentic video generation framework that combines causal reasoning over the observed physical failure with prior repair experience to enable localized and adaptive correction. Specifically, Spatiotemporal Causal Constraints identify the first invalid physical state and its causally affected states, producing a repair specification that guides physically plausible local regeneration while preserving unaffected content. Experience-augmented In-context Learning combines the ongoing repair context with relevant experience from previous repairs, allowing the agent to adapt its repair decisions across diverse physical failures and accumulate completed trajectories for future cases. Experiments on Physics-IQ Verified across solid mechanics, fluid dynamics, optics, thermodynamics, and magnetism demonstrate that PhysRepair consistently improves the physical plausibility of generated videos over existing methods. Our project page is available at https://xxkkxxx.github.io/PhysRepair-Website/.