Ključne besede: diffusion models, text-to-image generation, human body segmentation, inpainting, virtual try-onCelotno besedilo (datoteka, 7,22 MB)