SynMotion：面向运动定制视频生成的语义—视觉自适应
SynMotion: Semantic-Visual Adaptation for Motion Customized Video Generation
把主体与运动表征解耦成双嵌入、再叠加参数高效运动适配器的做法，可直接迁移到任何需要「保留主体身份、只换动作」的视频扩散定制任务；配套的交替训练策略与 SPV 数据构造方式也为少样本运动定制提供了可复用的训练范式。
原文依据 we integrate parameter-efficient motion adapters into a pre-trained video generation model to enhance motion fidelity and temporal coherence
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论基于扩散的视频运动定制能够从少量视频样本中获取人体运动表征，同时通过精确的文本条件实现任意主体的迁移。现有方法通常依赖语义层面的对齐，期望模型学习新的运动概念并将其与其他实体（如猫或狗）组合，从而生成视觉上吸引人的结果。然而，视频数据涉及复杂的时空模式，仅关注语义会导致模型忽略运动的视觉复杂性；反之，仅微调视觉表征又会在表示目标动作时产生语义混淆。为解决上述局限，我们提出 SynMotion，一种联合利用语义引导与视觉自适应的新运动定制视频生成模型。在语义层面，我们引入双嵌入语义理解机制，将主体表征与运动表征解耦，使模型在学习定制运动特征的同时保留其对多样主体的生成能力。在视觉层面，我们将参数高效的运动适配器集成到预训练视频生成模型中，以增强运动保真度与时序连贯性。此外，我们提出一种新的嵌入专属训练策略，交替优化主体嵌入与运动嵌入，并辅以人工构建的主体先验视频（SPV）训练数据集。该策略在提升运动特异性的同时保持对多样主体的泛化能力。最后，我们推出 MotionBench，一个涵盖多样运动模式的新构建基准。在 T2V 与 I2V 两种设定下的实验结果表明，SynMotion 优于现有基线。
英文原文摘要 1790 字符
Diffusion-based video motion customization facilitates the acquisition of human motion representations from a few video samples, while achieving arbitrary subjects transfer through precise textual conditioning. Existing approaches often rely on semantic-level alignment, expecting the model to learn new motion concepts and combine them with other entities (e.g., cats or dogs) to produce visually appealing results. However, video data involve complex spatio-temporal patterns, and focusing solely on semantics cause the model to overlook the visual complexity of motion. Conversely, tuning only the visual representation leads to semantic confusion in representing the intended action. To address these limitations, we propose SynMotion, a new motion-customized video generation model that jointly leverages semantic guidance and visual adaptation. At the semantic level, we introduce the dual-embedding semantic comprehension mechanism which disentangles subject and motion representations, allowing the model to learn customized motion features while preserving its generative capabilities for diverse subjects. At the visual level, we integrate parameter-efficient motion adapters into a pre-trained video generation model to enhance motion fidelity and temporal coherence. Furthermore, we introduce a new embedding-specific training strategy which alternately optimizes subject and motion embeddings, supported by the manually constructed Subject Prior Video (SPV) training dataset. This strategy promotes motion specificity while preserving generalization across diverse subjects. Lastly, we introduce MotionBench, a newly curated benchmark with diverse motion patterns. Experimental results across both T2V and I2V settings demonstrate that SynMotion outperforms existing baselines.