Reproduce “StereoGaussians: Feed-Forward 3D Gaussian Splatting from Stereo Images” 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
Boyuan Tian, Huangying Zhan, Zhan Li, Shin-Fang Chng, Hanwen Yang, Zirui Wang, Yi Xu
StereoGaussians converts a calibrated stereo pair into a renderable Gaussian scene, using frozen stereo features, metric depth for visible geometry, a back layer, and a padded canvas. On two zero-shot benchmarks it leads evaluated methods; on ReplicaGS its PSNR is 30.266 dB versus LVSM’s 25.772 dB. Evidence supports tested nearby views, not large camera travel or perfect recovery of hidden surfaces.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.