MesonGS++: Post-training Compression of 3D Gaussian Splatting with Hyperparameter Searching

2026-04-29

AI Summary

This work addresses the large storage overhead of 3D Gaussian Splatting (3DGS) models and the limitations of existing post-training compression methods, which rely on coupled hyperparameters and struggle to precisely control compression ratios while optimizing rate-distortion trade-offs. The authors propose a target-budget-driven post-training compression framework that jointly optimizes Gaussian retention rates and bit-width allocation, achieving—for the first time—exact adherence to a specified compression ratio while simultaneously improving rendering quality. Their approach introduces a size-aware hyperparameter search mechanism combining a linear size estimator with integer programming, and integrates importance-aware pruning, octree-based geometry encoding, selective vector quantization, grouped mixed-precision quantization, and CUDA-optimized parallel operators. Experiments demonstrate 20–34× compression ratios with maintained or even enhanced PSNR, and in some scenes, the compressed model at 20× outperforms the original 3DGS in rendering quality.

Abstract

3D Gaussian Splatting (3DGS) achieves high-quality novel view synthesis with real-time rendering, but its storage cost remains prohibitive for practical deployment. Existing post-training compression methods still rely on many coupled hyperparameters across pruning, transformation, quantization, and entropy coding, making it difficult to control the final compressed size and fully exploit the rate-distortion trade-off. We propose MesonGS++, a size-aware post-training codec for 3D Gaussian compression. On the codec side, MesonGS++ combines joint importance-based pruning, octree geometry coding, attribute transformation, selective vector quantization for higher-degree spherical harmonics, and group-wise mixed-precision quantization with entropy coding. On the configuration side, it treats the reserve ratio and bit-width allocation as the dominant rate-distortion knobs and jointly optimizes them under a target storage budget via discrete sampling and 0--1 integer linear programming. We further propose a linear size estimator and a CUDA parallel quantization operator to accelerate the hyperparameter searching process. Extensive experiments show that MesonGS++ achieves over 34$\times$ compression while preserving rendering fidelity, outperforming state-of-the-art post-training methods and accurately meeting target size budgets. Remarkably, without any training, MesonGS++ can even surpass the PSNR of vanilla 3DGS at a 20$\times$ compression rate on the Stump scene. Our code is available at https://github.com/mmlab-sigs/mesongs_plus

Problem

• 3D Gaussian Splatting
• post-training compression
• rate-distortion trade-off
• storage cost
• hyperparameter coupling

Innovation

• post-training compression
• 3D Gaussian Splatting
• rate-distortion optimization
• mixed-precision quantization
• hyperparameter search