Reproduce “DeepVoxels: Learning Persistent 3D Feature Embeddings” 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
Vincent Sitzmann, Justus Thies, Felix Heide, Matthias Nießner, Gordon Wetzstein, Michael Zollhöfer
DeepVoxels stores each scene in a persistent 3D grid of features and renders it from requested camera views using camera geometry and learned soft visibility for occlusion. It learns from posed 2D images without 3D geometry supervision. Across four synthetic objects, it averages 30.55 PSNR and 0.97 SSIM, versus Pix2Pix’s 23.63 and 0.92; results cover tested views, not arbitrary viewpoints.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.