Reproduce “ChronoFuseGS: Multi-Temporal Gaussian Fusion with Per-Splat Persistence and Change Visualization” 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

Tobias Batik, Diana Marin, Peter Kán, Hannes Kaufmann

ChronoFuseGS merges separately trained 3D Gaussian Splatting models from repeated captures, assigning each Gaussian per-timestep persistence and refining the combined model across images. This transfers supported detail between scene states and enables Gaussian-level change visualization. On the NWM Complete subset, it reached 21.9 PSNR, 0.74 SSIM, and 0.43 LPIPS, with small whole-image gains and assumptions about registration and within-period stability.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.