Reproduce “Birth-Death Structural Learning 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
Quan Tran Hong, Long Nguyen-Chi, Binh T. Nguyen
The authors present a controller for 3D Gaussian Splatting that allocates a Gaussian budget using primitives’ first-order effects on image loss. It scores opacity-mass changes at existing Gaussians to guide births and low-opacity deaths. On seven Mip-NeRF360 scenes at 100k Gaussians, it raises test-view PSNR from 27.61 to 27.94 dB versus a matched opacity controller, but gains are not uniform across budgets.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.