Reproduce “Surface Reconstruction from Range Scans Under Visibility Constraints” 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

Gabriel Lenain, Edmond Boyer, Jean Ponce

The authors reconstruct surfaces from range scans by matching each measured depth to the first predicted surface hit along its sensor ray. A 1-Lipschitz implicit field enables sphere tracing, while a Huber loss limits large residuals. On the Surface Reconstruction Benchmark with 20% outliers, the L=2 model achieves CD₁ 0.122 ± 0.004, but finite rays cannot uniquely determine unseen geometry.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.