Fast Converging 3D Gaussian Splatting for 1-Minute Reconstruction

2026-01-27

AI Summary

This work proposes a two-stage adaptive 3D Gaussian Splatting (3DGS) framework for rapid high-quality 3D scene reconstruction within one minute, accommodating heterogeneous camera pose inputs—noisy SLAM poses and accurate COLMAP poses. In the first stage, tailored for noisy SLAM trajectories, the method introduces anchor-based neural Gaussian representations, reverse parallel optimization, global pose refinement, and monocular depth initialization. The second stage transitions to standard 3DGS optimized for precise COLMAP poses, incorporating multi-view consistency-guided Gaussian splitting and depth supervision to eliminate MLP overhead and enhance geometric fidelity. The approach achieved a PSNR of 28.43, winning the SIGGRAPH Asia 3DGS Fast Reconstruction Challenge by effectively balancing reconstruction speed and visual accuracy.

Technology Category

• Computer Vision: 3D Computer Vision
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Search and Optimization: Sampling/Simulation-based Search

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: Data quality aspects of human-annotated datasets

Abstract

We present a fast 3DGS reconstruction pipeline designed to converge within one minute, developed for the SIGGRAPH Asia 3DGS Fast Reconstruction Challenge. The challenge consists of an initial round using SLAM-generated camera poses (with noisy trajectories) and a final round using COLMAP poses (highly accurate). To robustly handle these heterogeneous settings, we develop a two-stage solution. In the first round, we use reverse per-Gaussian parallel optimization and compact forward splatting based on Taming-GS and Speedy-splat, load-balanced tiling, an anchor-based Neural-Gaussian representation enabling rapid convergence with fewer learnable parameters, initialization from monocular depth and partially from feed-forward 3DGS models, and a global pose refinement module for noisy SLAM trajectories. In the final round, the accurate COLMAP poses change the optimization landscape; we disable pose refinement, revert from Neural-Gaussians back to standard 3DGS to eliminate MLP inference overhead, introduce multi-view consistency-guided Gaussian splitting inspired by Fast-GS, and introduce a depth estimator to supervise the rendered depth. Together, these techniques enable high-fidelity reconstruction under a strict one-minute budget. Our method achieved the top performance with a PSNR of 28.43 and ranked first in the competition.

Problem

• 3D Gaussian Splatting
• fast reconstruction
• one-minute budget
• camera pose noise
• real-time 3D reconstruction

Innovation

• 3D Gaussian Splatting
• fast reconstruction
• Neural-Gaussian representation
• pose refinement
• multi-view consistency