Reproduce “Scene Representation Networks: Continuous 3D-Structure-Aware Neural Scene Representations” 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, Michael Zollhöfer, Gordon Wetzstein
SRNs represent objects as continuous 3D fields and render views by tracing camera rays. A learned field maps world coordinates to features; ray marching estimates surfaces, and a per-pixel network predicts color. On held-out synthetic ShapeNet chairs from two images, they report 24.48 dB PSNR and 0.92 SSIM, versus 22.28 dB and 0.90 for WRL, in this controlled object setting.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.