SIFT: Self-Imagination Fine-Tuning for Physically Plausible Motion in Video Diffusion Models figure
AlphaXiv 中文概览(可滚动查看)