Reproduce “Counting Contributions, Not Views: Occlusion-Aware Densification for 3D Gaussian Splatting” 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
Marcel Rogge, René Schuster, Didier Stricker
The authors modify 3D Gaussian Splatting densification for scenes with occlusion. Their filter counts a camera view only when a Gaussian contributes to a rendered pixel, and their criterion scales gradient signals by mean pixel contribution. On four Zip-NeRF apartment scenes, the method improves PSNR, SSIM, LPIPS, and GMSD over PixelGS with distance scaling, using 21% fewer Gaussians; this comparison does not establish lower computation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.