Shigon/BRDFusion_videos

BRDFusion Evaluation Videos

This repository contains the evaluation videos for the paper BRDFusion: Physics Meets Generation for Urban Scene Inverse Rendering.

Project Page | GitHub | arXiv | Paper

BRDFusion is a unified framework that combines physics-based inverse rendering with generative modeling for high-quality urban scene relighting. It decomposes urban videos into geometry, materials, and HDR lighting for novel-view rendering, relighting, and scene-edit applications.

Sample Usage

To evaluate these precomputed videos and compare against baselines (such as UrbanIR, InvRGB+L, and Gen3C+DR), you can use the evaluation script provided in the official GitHub repository.

After cloning the repository and setting up the brdfusion environment, run the following command:

conda run -n brdfusion --no-capture-output \
  python -u tools/metrics/compute_precomputed_video_metrics.py --root ./videos/

Video Structure

The precomputed videos are organized by method and dataset:

videos/
  brdfusion/{self,waymo}/...
  gen3c_dr/{self,waymo}/...
  invrgbl/{self,waymo}/...
  urbanir/{self,waymo}/...

Citation

@misc{liu2026brdfusionphysicsmeetsgeneration,
  title={BRDFusion: Physics Meets Generation for Urban Scene Inverse Rendering}, 
  author={Yi-Ruei Liu and Jie-Ying Lee and Zheng-Hui Huang and Yu-Lun Liu and Chih-Hao Lin},
  year={2026},
  eprint={2606.17049},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2606.17049}, 
}