Reproduce “M-plicits: Neural Implicit Surfaces via Nested Multiscale Residuals” 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

Vinícius da Silva, Isabelle Melo, Matheus Bessa, Guilherme Schardong, Luiz Schirmer, André Araújo, Nuno Gonçalves, Hélio Lopes and 3 more

M-plicits reconstructs 3D surfaces from oriented points with a zero-level signed distance function. It trains a coarse network, then adds residual networks supervised in narrow bands around the preceding surface, including input points. With up to 1% vertex noise, fine M-plicits has mean CD of 3.36×10⁻⁵, versus 2.80×10⁻⁴ for fine iNGP; this aggregate result does not establish wins on every shape or metric.

Summary by alphaXiv.

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