FlexGaussian: Flexible and Cost-Effective Training-Free Compression for 3D Gaussian Splatting

摘要

3D Gaussian Splatting has emerged as a prominent technique for representing and rendering complex 3D scenes, offering high fidelity and speed but resulting in large file sizes. Existing compression methods can reduce 3D Gaussian data size but often require costly retraining or refinement, which is memory- and compute-intensive and lacks flexibility for varying compression needs. This challenge grows as large-scale scenes become more common, increasing the demand for efficient, low-overhead compression methods - especially for resource-constrained mobile and edge devices. We introduce FlexGaussian, a flexible, cost-effective, training-free compression method for 3D Gaussian Splatting that combines mixed-precision quantization with attribute-discriminative pruning. FlexGaussian requires no retraining or refinement and adapts easily to diverse compression targets. Evaluations demonstrate up to 96.4% compression with minimal rendering quality loss (<1 dB PSNR drop) and support for mobile deployment. FlexGaussian achieves high compression ratios within seconds, running 1.7-2.1× faster than state-of-the-art training-free methods and 10-100×faster than training-involved approaches. Project page: https://supercomputing-system-ai-lab.github.io/projects/flexgaussian.