Reproduce “Semantic Image Synthesis via Diffusion Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Wengang Zhou, Weilun Wang, Jianmin Bao, Dongdong Chen, Dong Chen, Lu Yuan, Houqiang Li
Semantic Diffusion Model (SDM) synthesizes photographs from semantic layouts by progressively denoising noise, routing the noisy image through a U-Net encoder and the map into the decoder through spatially adaptive normalization. Classifier-free guidance strengthens the map’s influence during sampling. On Cityscapes, SDM has lower FID but lower LPIPS than ControlNet; trained from scratch, it does not generalize to natural images.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.