GS-ROR²: Bidirectional-guided 3DGS and SDF for Reflective Object Relighting and Reconstruction

This repository contains the official implementation of the following paper:

GS-ROR²: Bidirectional-guided 3DGS and SDF for Reflective Object Relighting and Reconstruction
Zuo-Liang Zhu¹, Beibei Wang², Jian Yang^1
1Nankai University ²Nanjing University
In ACM TOG 2025

[Paper] [Project Page] [Video (TBD)]

GS-ROR² is an efficient, robust solution for object relighting and reconstruction, aiming to produce decent decompositions of geometry, material, and lighting along with high-quality meshes from multi-view observations.

News

• Apr. 28, 2025: Our code is publicly available.
• Aug. 6, 2025: Our paper is publicly available on ArXiv.
• Aug. 6, 2025: Release pretrained models.

Method Overview

For more technical details, please refer to our paper on arXiv.

Dependencies and Installation

1. Clone repo.

   git clone https://github.com/NK-CS-ZZL/GS-ROR.git
   cd GSROR

2. Create Conda environment and install dependencies

   conda create -n gsror python=3.10
   conda activate gsror
   pip install torch==2.2.1 torchvision==0.17.1 --index-url https://download.pytorch.org/whl/cu118
   pip install -r requirements.txt
   git clone https://github.com/NVlabs/nvdiffrast
   pip install ./nvdiffrast
   pip install ./submodules/fused-ssim
   pip install ./submodules/r3dg-rasterization
   pip install ./submodules/simple-knn

   Note that

   • Our code is verfied under CUDA11.8 runtime, so we recommend to use the same environment to guarantee reproductibility.
   • Please switch to the corresponding runtime if the NVCC version is higher than 11.8.

3. Download pretrained models for demos from Download and place them to the pretrained folder

Download

Dataset	Training Bash	🔗 Source	🔗 Checkpoint	🔗 Result
Glossy Synthetic	train_glossy.sh	Images	HuggingFace	HuggingFace
Shiny Blender	train_shiny.sh	Images / Point Cloud	HuggingFace	HuggingFace
TensoIR Synthetic	train_tir.sh	Images / Env. maps	HuggingFace	HuggingFace

Training and Evaluation

Please refer to development.md to learn how to benchmark the DiscretizedSDF and how to train yourself DiscretizedSDF model from the scratch.

Citation

If you find our repo useful for your research, please consider citing our paper:

@inproceedings{zhu_2025_gsror,
   title={GS-ROR^2: Bidirectional-guided 3DGS and SDF for Reflective Object Relighting and Reconstruction},
   author={Zhu, Zuo-Liang and Yang, Jian and Wang, Beibei},
   journal = {ACM Transactions on Graphics (TOG)},
   year={2025}，
   publisher = {ACM},
   doi={10.1145/3759248},
}

License

This code is licensed under the Creative Commons Attribution-NonCommercial 4.0 International for non-commercial use only. Please note that any commercial use of this code requires formal permission prior to use.

Contact

For technical questions, please contact nkuzhuzl[AT]gmail.com.

For commercial licensing, please contact beibei.wang[AT]nju.edu.cn。

Acknowledgement

We thank Jia Li for his suggestions during the project.

Here are some great resources we benefit from: GaussianShader, NeRO, TensoSDF, Ref-NeuS, R3DG

If you develop/use GS-ROR in your projects, welcome to let us know. We will list your projects in this repository.