SCP-Diff: Photo-Realistic Semantic Image Synthesis with Spatial-Categorical Joint Prior

arXiv:2403.09638v1 Announce Type: new
Abstract: Semantic image synthesis (SIS) shows good promises for sensor simulation. However, current best practices in this field, based on GANs, have not yet reached the desired level of quality. As latent diffusion models make significant strides in image generation, we are prompted to evaluate ControlNet, a notable method for its dense control capabilities. Our investigation uncovered two primary issues with its results: the presence of weird sub-structures within large semantic areas and the misalignment of …

abstract arxiv best practices categorical controlnet cs.cv current diff diffusion diffusion models gans good however image image generation latent diffusion models photo practices prior quality semantic sensor shows simulation spatial synthesis type