Abstract: We seek to give users precise control over diffusion-based image generation by modeling complex scenes as sequences of layers, which define the desired spatial arrangement and visual ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results