3D scenes simplified drastically for faster real-time rendering
CVT-GS: Learning to Simplify 3D Gaussian Splatting with Centroidal Voronoi Tessellation
Computer Vision and Pattern Recognition
Summary
Creating detailed 3D scenes for virtual viewing usually needs many building blocks called Gaussian primitives, which slows down the process and takes up a lot of memory. The researchers developed a way to reduce the number of these blocks without losing picture quality, by cleverly grouping nearby ones and merging their details into a single representative block. Their method works after a scene is already made, so it doesn’t require changing how the scenes are created or extra training. This approach makes rendering much faster and uses less storage while even improving the quality compared to previous methods.
3D Gaussian Splattingnovel view synthesisGaussian primitivesCentroidal Voronoi Tessellationdifferentiable renderingscene simplificationpost-hoc compressionPSNRneural cell merger
Authors
Bingxian Li, Yilong Li, Jingliang Peng, Peng-Shuai Wang, Fei Zhu, Guozheng Li, Chi Harold Liu, Guoping Wang, Bo Pang
Abstract
While 3D Gaussian Splatting (3DGS) has emerged as a powerful representation for real-time novel view synthesis, rendering high-fidelity scenes often relies on a massive number of Gaussian primitives, incurring substantial storage and computational overhead. Existing simplification techniques are largely intrusive, requiring training-time pruning, architectural modifications, or computationally expensive per-scene fine-tuning. These drawbacks limit their deployment on off-the-shelf pretrained models. In this paper, we propose CVT-GS, a novel optimization-free post-hoc simplification framework that directly compresses trained 3DGS scenes without sacrificing visual fidelity. Our approach first constructs spatially coherent cells over Gaussian centers via a geometry-aware Centroidal Voronoi Tessellation (CVT). Subsequently, a lightweight neural cell merger predicts the geometry and appearance of a single, highly representative Gaussian primitive for each cell under differentiable rendering supervision. By formulating simplification as a rendering-aware many-to-one merging process rather than naive primitive pruning, CVT-GS outputs a standard 3DGS scene that is seamlessly compatible with existing renderers. Experiments on various datasets demonstrate the superiority of our method. Notably, when achieving a 100-fold reduction in Gaussian points, our method operates 12 times faster than state-of-the-art methods while improving the PSNR by 1.3 dB.