REFINE: Super-efficient 3D Gaussian Splatting Pruning via Rendering-Free Primitive Importance
2026-06-08 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors present REFINE, a new method to make 3D Gaussian splatting models smaller and faster without losing much quality. Instead of rendering the scene many times to decide which parts to remove, their method uses math to estimate how important each piece is, saving a lot of computation. They tested REFINE on several datasets and found it keeps good image quality while being about 3,000 times faster than existing methods. This makes pruning 3D models much more practical.
3D Gaussian SplattingPruningRenderingHessian FieldPerceptual ErrorVisibilityProjection GeometryComputational ComplexityAdaptive Hyperparameter
Authors
Zhang Chen, Shuai Wan, Mengting Yu, Fuzheng Yang, Junhui Hou
Abstract
Existing pruning methods for 3D Gaussian splatting (3DGS) suffer from either severe quality degradation or prohibitive computational overhead. In this paper, we propose REFINE, a highly accelerated 3DGS pruning framework centered on a novel rendering-free primitive importance metric. Our approach leverages an analytically approximated, rendering-aware Hessian field to quantify the expected perceptual error induced by the removal of individual primitives. By modeling the joint modulation of visibility, projection geometry and the content adaptive hyperparameter, we entirely bypass costly forward rendering passes and derive an anisotropic perceptual weight field that serves as a high-fidelity proxy for primitive importance. Extensive experiments across multiple benchmark datasets demonstrate that REFINE maintains highly competitive rendering quality while achieving an unprecedented $3,000\times$ reduction in pruning-related computational complexity compared to state-of-the-art pruning methods.