ReImagine: Rethinking Controllable High-Quality Human Video Generation via Image-First Synthesis
This work addresses the challenge of controllable human video generation by introducing an image-first paradigm that improves visual quality and temporal consistency, though it is an incremental step over existing methods.
Human video generation is reformulated as an image-first synthesis problem, decoupling appearance from temporal modeling. The proposed pipeline achieves high-quality, temporally consistent videos under diverse poses and viewpoints, with a training-free temporal refinement stage.
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 videos under diverse poses and viewpoints. We also release a canonical human dataset and an auxiliary model for compositional human image synthesis. Code and data are publicly available at https://github.com/Taited/ReImagine.