Reproduce “Morphable Diffusion: 3D-Consistent Diffusion for Single-image Avatar Creation” 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
Xiyi Chen, Marko Mihajlovic, Shaofei Wang, Sergey Prokudin, Siyu Tang
Morphable Diffusion generates face views from one image using a mesh that specifies a target expression. During denoising, it links image features across views to mesh vertices, while a diffusion model supplies appearance. On FaceScape’s held-out “jaw right” test, it reports [email protected] of 95.46, versus 89.74 for personalized DiffusionRig*, within restricted camera angles; this supports better keypoint alignment in that protocol.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.