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

TL;DR

MesonGS++ is a post-training compression method for 3D Gaussian Splatting that significantly reduces storage costs while maintaining high rendering quality, using hyperparameter optimization and advanced coding techniques.

Contribution

It introduces a size-aware codec with joint hyperparameter optimization and acceleration techniques, outperforming existing post-training compression methods for 3D Gaussian Splatting.

Findings

Achieves over 34× compression while preserving fidelity.

Surpasses vanilla 3DGS PSNR at 20× compression without training.

Outperforms state-of-the-art post-training compression methods.

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