Keywords: diffusion models, text-to-image generation, human body segmentation, inpainting, virtual try-onFull text (file, 7,22 MB)