[论文] ReImagine: Rethinking Controllable High-Quality Human Video Generation...

## 论文概要 **研究领域**: CV **作者**: Zhengwentai Sun, Keru Zheng, Chenghong Li, Hongjie Liao, Xihe Yang, Heyuan Li, Yihao Zhi, Shuliang Ning, Shuguang Cui, Xiaoguang Han **发布时间**: 2026-04-21 **arXiv**: [2604.19720](https://arxiv.org/abs/2604.19720) ## 中文摘要 人体视频生成仍具挑战性，因难以在有限多视角数据下联合建模人体外观、运动与相机视角。现有方法常分别处理这些因素，导致可控性有限或视觉质量下降。我们从图像优先视角重新审视此问题：通过图像生成学习高质量人体外观并用作视频合成先验，解耦外观建模与时序一致性。我们提出姿态与视角可控管道，结合预训练图像骨干与基于 SMPL-X 的运动引导，以及基于预训练视频扩散模型的免训练时序精化阶段。我们的方法在多样姿态与视角下生成高质量、时序一致的视频。我们还发布规范人体数据集与组合人体图像合成的辅助模型。代码与数据见 https://github.com/Taited/ReImagine。 ## 原文摘要 Human video generation remains challenging due to the difficulty of jointly modeling human appearance, motion, and camera viewpoint under limited multi-view data. Existing methods often address these factors separately, resulting in limited controllability or reduced visual quality. We revisit this problem from an image-first perspective, where high-quality human appearance is learned via image generation and used as a prior for video synthesis, decoupling appearance modeling from temporal consistency. We propose a pose- and viewpoint-controllable pipeline that combines a pretrained image backbone with SMPL-X-based motion guidance, together with a training-free temporal refinement stage based on a pretrained video diffusion model. Our method produces high-quality, temporally consistent vid... --- *自动采集于 2026-04-23* #论文 #arXiv #CV #小凯