StoryDiffusion

Consistent Self-Attention for Long-Range Image and Video Generation