外观
论文HF Daily Papers · arXiv 2610.299157 天前
CausalAlign框架提升视频生成自运动一致性，具身任务成功率增10%
本文由 AI 根据公开内容整理，可能有误，请以原文为准：CausalAlign: Causal Alignment of Video Generation Models for Consistent Ego-Motion（HF Daily Papers · arXiv 2610.29915）
核心内容
针对扩散视频生成模型（VGMs）长时序相机自运动漂移痛点，研究提出CausalAlign因果对齐框架。该方法基于Wan2.2模型在DL3DV和RealEstate10K数据集微调，实测几何一致性较基座模型提升超20%，且在RoboCasa365基准测试中，使世界动作模型（WAMs）的移动操作任务成功率提升10个百分点，验证了其对具身智能下游任务的实质增益。
技术层面，研究利用线性探针与因果干预诊断出VGM去噪过程中自运动表征呈“升-降”模式，据此定位关键层并通过Delta-Token空间将其与3D基础模型对齐。该空间编码连续帧间变化量，迫使模型聚焦自运动动态而非静态场景内容，从而在保持视觉质量的同时，显著增强运动精度、几何一致性与时间平滑度，为具身仿真数据生成提供可靠方案。
关键看点
- 对齐机制：构建Delta-Token空间编码帧间变化量，将VGM关键层表征与3D基础模型对齐，聚焦自运动动态而非静态内容
- 性能指标：Wan2.2微调后几何一致性相对基座提升超20%，RoboCasa365移动操作任务成功率增加10个百分点
- 诊断发现：通过线性探针与因果干预揭示VGM去噪过程中自运动状态表征存在特定的“升-降”演化规律
工程与产业观察
该工作证实视频生成模型的自运动一致性是具身策略学习的关键瓶颈。通过结构化对齐3D几何先验，可将通用视频模型转化为高保真具身仿真器，直接提升WAM等下游模型在复杂操作任务中的泛化能力，为合成数据驱动的具身训练提供了可复现的工程范式。
英文摘要 (Abstract)
Diffusion-based video generation models (VGMs) can produce visually plausible videos, yet their camera ego-motion often drifts over long horizons. Camera ego-motion determines how the viewpoint changes between frames and is essential for maintaining consistent geometry and trajectories. To address this problem, we propose CausalAlign, a causal alignment framework with two components. First, we diagnose how precise ego motion is represented, used, and retained within VGMs using linear probes and causal interventions, revealing a rise-and-fall pattern during denoising. Second, we strengthen ego-motion state in VGM by aligning the representations at the identified layers to a 3D foundation model through delta-token space. This space encodes the changes between consecutive frames, allowing the alignment to focus on ego-motion dynamics rather than static scene content. With Wan2.2 fine-tuned on DL3DV and RealEstate10K, CausalAlign can largely improve motion accuracy, geometric consistency, and temporal smoothness, while preserving visual quality. For example, it improves geometric consistency by more than 20% relative to the base model. Moreover, with the fine-tuned VGM, CausalAlign improves the success rate of world action models (WAMs) by 10 points on mobile manipulation tasks of RoboCasa365 benchmark. Overall, CausalAlign improves ego-motion consistency in diffusion-based VGMs and yields noticeable gains in embodied models.
Hugging Face Daily Papers 热度：4 upvotes 作者：KaiZhu, Yintianrun Zhang, Shuwei Jiang, Ruolin Zuo, Yixin Chen, Duomin Wang, Miao Liu, Yin Li, Daquan Zhou, Yiwu Zhong
来源
本页事实对应的公开页面，以原文为准。