README.md

metadata

license: apache-2.0
pipeline_tag: image-to-3d
tags:
  - gaussian-splatting
  - 3d
  - surface-reconstruction

Revisiting Photometric Ambiguity for Accurate Gaussian-Splatting Surface Reconstruction

This repository provides the reconstructed meshes and resources for the paper Revisiting Photometric Ambiguity for Accurate Gaussian-Splatting Surface Reconstruction.

Authors: Jiahe Li, Jiawei Zhang, Xiao Bai, Jin Zheng, Xiaohan Yu, Lin Gu, Gim Hee Lee.

• 📚 Paper
• 🌐 Project Page
• 💻 Code

Overview

AmbiSuR is a framework that explores an intrinsic solution upon Gaussian Splatting for photometric ambiguity-robust surface 3D reconstruction. By revisiting built-in primitive-wise ambiguities, the framework introduces a photometric disambiguation constraint and an ambiguity indication module to identify and guide the correction of underconstrained reconstructions, achieving high-performance surface formation in challenging scenarios.

Reconstruction on Tanks and Temples and DTU Datasets

Here we provide the reconstructed meshes of the paper's experiments from AmbiSuR.

You can browse all the released meshes at:

• ambisur-<dataset>-meshes-eval/: The meshes on DTU/TnT datasets, with strict filtering strategy for evaluation.
• ambisur-<dataset>-meshes-vis/: The meshes on DTU/TnT datasets, with loose filtering strategy for visualization.

Metrics shall be reproduced with the results with postfix of -eval.

Download

from huggingface_hub import snapshot_download
snapshot_download(repo_id="Fictionary/AmbiSuR", cache_dir='./AmbiSuR/results', local_dir ='./AmbiSuR/results')

or use Git to clone this repository with LFS.

Citation

@inproceedings{li2026ambisur,
  title={Revisiting Photometric Ambiguity for Accurate Gaussian-Splatting Surface Reconstruction},
  author={Li, Jiahe and Zhang, Jiawei and Bai, Xiao and Zheng, Jin and Yu, Xiaohan and Gu, Lin and Lee, Gim Hee},
  booktitle={International Conference on Machine Learning},
  year={2026},
  organization={PMLR}
}