arXiv cs.AI· Yuxuan Chen, Xiaojie Li, Hao Chen, Juncheng Ma, Kewei Zhang, Zhongjing Du, Guoqiang Wei, Fangfang Wang, Zhen Dong, Zhiyang Dou, Liming Zhao, Yan Zeng, Jianwei Zhang, Daquan Zhou·· 1 天前AI 评分42
PhysLoop：利用 LLM 物理认知实现视频生成模型自我改进
PhysLoop: Harnessing LLMs' Physics Awareness for Self-Improving Video Generation Models
AI 导读
PhysLoop 是一套统一框架，引导 LLM 在视频生成、筛选与修正全流程中提供物理监督，将物理知识蒸馏为物理一致的视频，并支持可控训练数据构建与推理时视频修复。团队开源了 10 万条覆盖室内外场景的视频、生成工具及 PhysLoop-Gen 与 PhysLoop-Repair 两个评测数据集，并在 Physics-IQ Verified 基准上评估物理表现。
正文 · 原文
Authors:Yuxuan Chen, Xiaojie Li, Hao Chen, Juncheng Ma, Kewei Zhang, Zhongjing Du, Guoqiang Wei, Fangfang Wang, Zhen Dong, Zhiyang Dou, Liming Zhao, Yan Zeng, Jianwei Zhang, Daquan Zhou
Abstract:Video generation models produce visually realistic videos but still struggle with physical dynamics and causal consistency. Learning from generated videos requires fine-grained physical supervision, yet reliable quality judgments, correction pairs, and controlled variations are difficult to obtain. Large language models (LLMs) possess physical knowledge, but struggle to reliably guide video generation models in producing their training videos. We present PhysLoop, a unified harness designed to support physically grounded self-improvement in video generation models. The harness guides and constrains LLMs throughout video generation, selection, and correction, distilling their physical knowledge into physically consistent videos. The framework supports controllable training-data construction and inference-time video repair. By identifying and correcting failures, PhysLoop marks a key milestone toward recursive self-improvement (RSI) in video generation models with limited human intervention. We release 100K videos spanning diverse, realistic indoor and outdoor scenes, along with generation tools and two evaluation datasets: PhysLoop-Gen and PhysLoop-Repair. We evaluate controllable generation and localized repair on these datasets, respectively, and assess physical behavior on the external Physics-IQ Verified benchmark.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.16748 [cs.CV] |
| (or arXiv:2611.16748v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2611.16748 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yuxuan Chen [view email]
[v1]
Thu, 12 Nov 2026 12:59:35 UTC (28,213 KB)
来源：arXiv cs.AI · arxiv.org