Reproduce “StereoSplat: Metric-Scale Novel View Synthesis via Stereo-Grounded 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
Vladimir Yugay, Denis Rozumny, Theo Gevers, Elias Vansteenkiste, Martin R. Oswald, Diogo C. Luvizon
StereoSplat reconstructs novel views from calibrated stereo video without per-scene optimization. It converts disparity from known stereo-camera geometry into metric depth, then uses depth and image features to place and aggregate 3D Gaussians. On KITTI, it improves depth accuracy over compared feed-forward systems and achieves 25.2 dB PSNR versus DepthSplat’s 22.1 dB; evaluation assumes static scenes and excludes dynamic clips.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.