视频扩散时间步中的运动编码表征
Characterizing Motion Encoding in Video Diffusion Timesteps
AI总结：
本文通过分析视频扩散模型中时间步的运动与外观编码特性，发现早期时间步主导运动，后期主导外观，并提出受限于运动主导阶段的运动定制方法。
摘要：
文本到视频扩散模型通过迭代去噪合成时间运动和空间外观，但如何在时间步之间编码运动仍缺乏深入理解。从业者常采用经验启发法，认为早期时间步主要塑造运动和布局，而后期时间步则细化外观，但这种行为尚未被系统性地表征。在本文中，我们通过在指定时间步范围内注入新条件时，出现编辑与运动保持之间的权衡来代理视频扩散时间步中的运动编码，并通过大规模定量研究表征这种代理。该协议使我们能够通过定量映射它们在去噪轨迹中的竞争关系来分离运动与外观。在多样化的架构中，我们一致地识别出一个早期以运动为主导的阶段和一个后期以外观为主导的阶段，从而在时间步空间中获得一个操作性的运动-外观边界。基于这种表征，我们通过将训练和推理限制在运动主导的阶段来简化当前的一次性运动定制范式，实现了强大的运动转移，而无需辅助去偏模块或专门的目标。我们的分析将一个广泛使用的启发法转化为一个时空解耦原则，我们的受时间步限制的配方可以作为现有运动转移和编辑方法的直接整合。
英文摘要：
Text-to-video diffusion models synthesize temporal motion and spatial appearance through iterative denoising, yet how motion is encoded across timesteps remains poorly understood. Practitioners often exploit the empirical heuristic that early timesteps mainly shape motion and layout while later ones refine appearance, but this behavior has not been systematically characterized. In this work, we proxy motion encoding in video diffusion timesteps by the trade-off between appearance editing and motion preservation induced when injecting new conditions over specified timestep ranges, and characterize this proxy through a large-scale quantitative study. This protocol allows us to factor motion from appearance by quantitatively mapping how they compete along the denoising trajectory. Across diverse architectures, we consistently identify an early, motion-dominant regime and a later, appearance-dominant regime, yielding an operational motion-appearance boundary in timestep space. Building on this characterization, we simplify current one-shot motion customization paradigm by restricting training and inference to the motion-dominant regime, achieving strong motion transfer without auxiliary debiasing modules or specialized objectives. Our analysis turns a widely used heuristic into a spatiotemporal disentanglement principle, and our timestep-constrained recipe can serve as ready integration into existing motion transfer and editing methods.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
