MuGS: Multi-Baseline Generalizable Gaussian Splatting Reconstruction

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Lou, Yaopeng, Shen, Liao, Liu, Tianqi, Li, Jiaqi, Huang, Zihao, Sun, Huiqiang, Cao, Zhiguo
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912667524923392
author Lou, Yaopeng
Shen, Liao
Liu, Tianqi
Li, Jiaqi
Huang, Zihao
Sun, Huiqiang
Cao, Zhiguo
author_facet Lou, Yaopeng
Shen, Liao
Liu, Tianqi
Li, Jiaqi
Huang, Zihao
Sun, Huiqiang
Cao, Zhiguo
contents We present Multi-Baseline Gaussian Splatting (MuGS), a generalized feed-forward approach for novel view synthesis that effectively handles diverse baseline settings, including sparse input views with both small and large baselines. Specifically, we integrate features from Multi-View Stereo (MVS) and Monocular Depth Estimation (MDE) to enhance feature representations for generalizable reconstruction. Next, We propose a projection-and-sampling mechanism for deep depth fusion, which constructs a fine probability volume to guide the regression of the feature map. Furthermore, We introduce a reference-view loss to improve geometry and optimization efficiency. We leverage 3D Gaussian representations to accelerate training and inference time while enhancing rendering quality. MuGS achieves state-of-the-art performance across multiple baseline settings and diverse scenarios ranging from simple objects (DTU) to complex indoor and outdoor scenes (RealEstate10K). We also demonstrate promising zero-shot performance on the LLFF and Mip-NeRF 360 datasets. Code is available at https://github.com/EuclidLou/MuGS.
format Preprint
id arxiv_https___arxiv_org_abs_2508_04297
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MuGS: Multi-Baseline Generalizable Gaussian Splatting Reconstruction
Lou, Yaopeng
Shen, Liao
Liu, Tianqi
Li, Jiaqi
Huang, Zihao
Sun, Huiqiang
Cao, Zhiguo
Computer Vision and Pattern Recognition
We present Multi-Baseline Gaussian Splatting (MuGS), a generalized feed-forward approach for novel view synthesis that effectively handles diverse baseline settings, including sparse input views with both small and large baselines. Specifically, we integrate features from Multi-View Stereo (MVS) and Monocular Depth Estimation (MDE) to enhance feature representations for generalizable reconstruction. Next, We propose a projection-and-sampling mechanism for deep depth fusion, which constructs a fine probability volume to guide the regression of the feature map. Furthermore, We introduce a reference-view loss to improve geometry and optimization efficiency. We leverage 3D Gaussian representations to accelerate training and inference time while enhancing rendering quality. MuGS achieves state-of-the-art performance across multiple baseline settings and diverse scenarios ranging from simple objects (DTU) to complex indoor and outdoor scenes (RealEstate10K). We also demonstrate promising zero-shot performance on the LLFF and Mip-NeRF 360 datasets. Code is available at https://github.com/EuclidLou/MuGS.
title MuGS: Multi-Baseline Generalizable Gaussian Splatting Reconstruction
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.04297