Reproduce “Budgeted-GS: Real-Time Large-Scale Gaussian Splatting via Factoring LOD” 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
Haipeng Wang
Budgeted-GS builds a post-training hierarchy for 3D Gaussian Splatting, merging primitives into aggregates and selecting detail by view with one budget control. On MatrixCity Aerial, the 2-GB-class tier uses 407 MB of model data and differs from the full render by at most 0.27 dB PSNR on a held-out view. Rendered on an RTX 4080 SUPER, it was not tested on target devices; the full-range law test failed on all 13 scenes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.