Search papers, labs, and topics across Lattice.
This paper introduces REFINE, a novel framework for pruning 3D Gaussian splatting (3DGS) that significantly reduces computational costs while preserving rendering quality. By utilizing a rendering-free primitive importance metric derived from an analytically approximated Hessian field, REFINE quantifies the perceptual error associated with removing primitives without the need for expensive forward rendering passes. Experimental results show that REFINE achieves a remarkable $3,000\times$ reduction in computational complexity compared to existing methods, while maintaining competitive rendering quality across various benchmark datasets.
Achieving a staggering $3,000\times$ reduction in computational complexity, REFINE redefines efficiency in 3D Gaussian splatting without sacrificing quality.
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.
