3D Gaussian splatting simplified for faster and smaller rendering
CVT-GS: Learning to Simplify 3D Gaussian Splatting with Centroidal Voronoi Tessellation
Computer Vision and Pattern Recognition
Summary
Rendering detailed 3D scenes often requires lots of tiny glowing spots called Gaussians, which can be slow and use a lot of storage. The authors propose a way to combine many of these spots into fewer, smarter ones without hurting image quality. They group nearby Gaussians using a special geometric method and then predict how to merge each group into one spot. This approach speeds up rendering significantly while keeping the scenes looking great, and it works on existing models without retraining.
What this means in practice
- •For 3d graphics developers: Reduce the complexity of 3D Gaussian scenes for faster rendering without retraining models.
- •For video game studios: Compress 3D environments to improve real-time rendering performance and lower storage requirements.
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.