FilterGS: Traversal-Free Parallel Filtering and Adaptive Shrinking for Large-Scale LoD 3D Gaussian Splatting

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Wang, Yixian, Yu, Haolin, Tang, Jiadong, Gao, Yu, Wang, Xihan, Yue, Yufeng, Yang, Yi
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917360988848128
author Wang, Yixian
Yu, Haolin
Tang, Jiadong
Gao, Yu
Wang, Xihan
Yue, Yufeng
Yang, Yi
author_facet Wang, Yixian
Yu, Haolin
Tang, Jiadong
Gao, Yu
Wang, Xihan
Yue, Yufeng
Yang, Yi
contents 3D Gaussian Splatting has revolutionized neural rendering with real-time performance. However, scaling this approach to large scenes using Level-of-Detail methods faces critical challenges: inefficient serial traversal consuming over 60\% of rendering time, and redundant Gaussian-tile pairs that incur unnecessary processing overhead. To address these limitations, we introduce FilterGS, featuring a parallel filtering mechanism with two complementary filters that select Gaussian elements efficiently without tree traversal. Additionally, we propose a novel GTC metric that quantifies the redundancy of Gaussian-tile key-value pairs. Based on this metric, we introduce a scene-adaptive Gaussian shrinking strategy that effectively reduces redundant pairs. Extensive experiments demonstrate that FilterGS achieves state-of-the-art rendering speeds while maintaining competitive visual quality across multiple large-scale datasets. Project page: https://github.com/xenon-w/FilterGS
format Preprint
id arxiv_https___arxiv_org_abs_2603_23891
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle FilterGS: Traversal-Free Parallel Filtering and Adaptive Shrinking for Large-Scale LoD 3D Gaussian Splatting
Wang, Yixian
Yu, Haolin
Tang, Jiadong
Gao, Yu
Wang, Xihan
Yue, Yufeng
Yang, Yi
Computer Vision and Pattern Recognition
3D Gaussian Splatting has revolutionized neural rendering with real-time performance. However, scaling this approach to large scenes using Level-of-Detail methods faces critical challenges: inefficient serial traversal consuming over 60\% of rendering time, and redundant Gaussian-tile pairs that incur unnecessary processing overhead. To address these limitations, we introduce FilterGS, featuring a parallel filtering mechanism with two complementary filters that select Gaussian elements efficiently without tree traversal. Additionally, we propose a novel GTC metric that quantifies the redundancy of Gaussian-tile key-value pairs. Based on this metric, we introduce a scene-adaptive Gaussian shrinking strategy that effectively reduces redundant pairs. Extensive experiments demonstrate that FilterGS achieves state-of-the-art rendering speeds while maintaining competitive visual quality across multiple large-scale datasets. Project page: https://github.com/xenon-w/FilterGS
title FilterGS: Traversal-Free Parallel Filtering and Adaptive Shrinking for Large-Scale LoD 3D Gaussian Splatting
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.23891