Improving Densification in 3D Gaussian Splatting for High-Fidelity Rendering

2025-08-17

AI Summary

3D Gaussian Splatting (3DGS) density optimization often induces geometric distortion and overfitting, degrading reconstruction quality. To address this, we propose a lightweight density optimization framework. First, we introduce an edge-aware scoring mechanism and a major-axis splitting strategy to precisely identify high-Gaussian-density regions and split Gaussians along their principal elongation directions—thereby mitigating shape distortion. Second, we integrate restoration-aware pruning, multi-step gradient updates, and growth control to jointly suppress overfitting. Crucially, all components incur no additional training or inference overhead. Evaluated on multiple benchmarks, our method achieves state-of-the-art rendering quality with significantly fewer Gaussians, improving both geometric reconstruction accuracy and visual fidelity. Without compromising computational efficiency, it establishes a new paradigm for high-fidelity, resource-efficient neural rendering.

Technology Category

• Computer Vision: 3D Computer Vision
• Search and Optimization: Distributed Search
• Machine Learning: Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Although 3D Gaussian Splatting (3DGS) has achieved impressive performance in real-time rendering, its densification strategy often results in suboptimal reconstruction quality. In this work, we present a comprehensive improvement to the densification pipeline of 3DGS from three perspectives: when to densify, how to densify, and how to mitigate overfitting. Specifically, we propose an Edge-Aware Score to effectively select candidate Gaussians for splitting. We further introduce a Long-Axis Split strategy that reduces geometric distortions introduced by clone and split operations. To address overfitting, we design a set of techniques, including Recovery-Aware Pruning, Multi-step Update, and Growth Control. Our method enhances rendering fidelity without introducing additional training or inference overhead, achieving state-of-the-art performance with fewer Gaussians.

Problem

• Optimizing densification strategy in 3D Gaussian Splatting
• Reducing geometric distortions during clone and split operations
• Mitigating overfitting to improve reconstruction quality

Innovation

• Edge-Aware Score for Gaussian selection
• Long-Axis Split reduces distortions
• Recovery-Aware Pruning prevents overfitting