发表机构
Seoul National University（首尔大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对视频生成中运动定制的内容泄漏问题，提出基于控制与随机最优控制的CMC框架，通过引导生成动态避免坍缩，提升训练效率2.5倍，兼顾运动保真与多样性。
AI 中文摘要
尽管视频生成领域近期取得了进展，运动定制仍然面临挑战，这主要源于内容泄漏问题，即参考视频中的外观属性会无意中传播到生成输出中。我们将此问题识别为生成过程向参考视频坍缩的结果，这一现象源于将学习目标设定为对参考视频的直接回归。为解决此问题，我们提出了基于控制的运动定制（CMC），这是一个在结构上对内容泄漏具有鲁棒性的原则性训练框架。我们的关键思想是引导生成动态朝向期望的运动，同时避免向参考视频坍缩，我们使用随机最优控制（SOC）对此进行形式化。在此公式下，定制视频获得目标运动，但仍保持在预训练模型的提示条件分布内，其中外观由文本提示而非参考视频决定。此外，为提高效率，我们针对运动定制调整了SOC公式，消除了对显式奖励的需求，并引入了仅关注早期生成阶段的时间步自适应运动成本，将训练速度提升了2.5倍。大量实验表明，CMC有效缓解了内容泄漏，在保持基础模型在多样场景下的多样性的同时，实现了具有竞争力的运动保真度。
英文摘要
Despite recent advances in video generation, motion customization remains challenging due to content leakage, where appearance attributes from the reference video unintentionally propagate into the generated output. We identify this issue as a consequence of the generative process collapsing toward the reference video, which arises from formulating the learning objective as a direct regression on the reference. To address this, we propose Control-based Motion Customization (CMC), a principled training framework that is structurally robust to content leakage. Our key idea is to steer generative dynamics toward desired motion while avoiding collapse toward the reference video, which we formalize using Stochastic Optimal Control (SOC). Under this formulation, customized videos acquire the target motion yet remain within the pre-trained model's prompt-conditional distribution, where appearance is determined by the text prompt rather than the reference video. Furthermore, to improve efficiency, we tailor the SOC formulation to motion customization by eliminating the need for an explicit reward and introducing a timestep-adaptive motion cost that focuses only on early generative stages, accelerating training by 2.5 times. Extensive experiments demonstrate that CMC effectively mitigates content leakage and achieves competitive motion fidelity while preserving the diversity of the base model across diverse scenarios.
CommentsPreprint
