Reproduce “Self-Consistency Geometry Injection for Multi-view Consistent Image Generation” 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
Yupeng Xiao, Yankai Wu, Jianjie Luo, Jiajie Lin, Yi Yu, Zhenguo Yang
RGMV guides multi-view image generation by scaling normal-conditioned signals according to agreement across repeated geometry predictions. Its reliability map is an unsupervised stability proxy, not a measure of geometric accuracy. On a fixed Objaverse subset, it improves PSNR, SSIM, and LPIPS over uniform injection with the RGB model frozen; on GSO, it also improves all three metrics over standalone Era3D.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.