CSS: Overcoming Pose and Scene Challenges in Crowd-Sourced 3D Gaussian Splatting

TL;DR

This paper presents CSS, a new 3D Gaussian Splatting pipeline that enables high-quality scene reconstruction from crowd-sourced images without known camera poses, overcoming traditional challenges of pose and scene variability.

Contribution

CSS introduces a robust 3D Gaussian Splatting method that handles pose-free crowd-sourced imagery for accurate scene reconstruction and novel view synthesis.

Findings

Improved scene reconstruction quality over existing methods.

Effective handling of pose-free crowd-sourced images.

Enhanced illumination modeling for realistic rendering.

Abstract

We introduce Crowd-Sourced Splatting (CSS), a novel 3D Gaussian Splatting (3DGS) pipeline designed to overcome the challenges of pose-free scene reconstruction using crowd-sourced imagery. The dream of reconstructing historically significant but inaccessible scenes from collections of photographs has long captivated researchers. However, traditional 3D techniques struggle with missing camera poses, limited viewpoints, and inconsistent lighting. CSS addresses these challenges through robust geometric priors and advanced illumination modeling, enabling high-quality novel view synthesis under complex, real-world conditions. Our method demonstrates clear improvements over existing approaches, paving the way for more accurate and flexible applications in AR, VR, and large-scale 3D reconstruction.