StoryDiffusion: Consistent Self-Attention for Long-Range Image and Video Generation Y upeng Zhou 1 Daquan Zhou 2 Ming-Ming Cheng