Tag
ReImagine introduces an image-first approach to controllable high-quality human video generation, combining SMPL-X motion guidance with video diffusion models to decouple appearance from temporal consistency.