Skip to content

[ICLR 2026] Official code for Dynamic Novel View Synthesis in High Dynamic Range which has been accepted by ICLR 2026.

License

Notifications You must be signed in to change notification settings

Surrey-UP-Lab/HDR-4DGS

Repository files navigation

Dynamic Novel View Synthesis in High Dynamic Range

Dynamic Novel View Synthesis in High Dynamic Range,
Kaixuan Zhang, Zhipeng Xiong, Minxian Li, Mingwu Ren, Jiankang Deng, Xiatian Zhu
Nanjing University of Science and Technology, Imperial College London and University of Surrey
ICLR 2026

This repository is the official implementation of "Dynamic Novel View Synthesis in High Dynamic Range". In this paper, we propose a more realistic problem named HDR Dynamic Novel View Synthesis (HDR DNVS), where the additional dimension “Dynamic” emphasizes the necessity of jointly modeling temporal radiance variations alongside sophisticated 3D translation between LDR and HDR. To tackle this complex, intertwined challenge, we introduce HDR-4DGS, a Gaussian Splatting-based architecture featured with an innovative dynamic tone-mapping module that explicitly connects HDR and LDR domains, maintaining temporal radiance coherence by dynamically adapting tone-mapping functions according to the evolving radiance distributions across the temporal dimension.

🛠️ Pipeline

Pipeline

Get started

Environment

The hardware and software requirements are the same as those of the 4D Gaussian Splatting, which this code is built upon. To setup the environment, please run the following command:

git clone https://github.com/prinasi/HDR-4DGS
cd HDR-4DGS
conda env create --file environment.yml
conda activate 4d_gaussian_splatting

Data preparation

HDR-4D-Syn dataset:

Download the proposed HDR-4D-Syn dataset and put each scene to data/HDR-4D-Syn.

HDR-4D-Real dataset:

The dataset can be downloaded from google drive. Then, put each scene into data/HDR-4D-Real.

Running

After the installation and data preparation, you can train the model by running:

chmod u+x ./single_train.sh
./single_train.sh <scene_name> <gpu_id>

📜 BibTex

@inproceedings{zhang2025dynamicnovelviewsynthesis,
  title={Dynamic Novel View Synthesis in High Dynamic Range},
  author={Kaixuan Zhang and Zhipeng Xiong and Minxian Li and Mingwu Ren and Jiankang Deng and Xiatian Zhu},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year={2026}
}

About

[ICLR 2026] Official code for Dynamic Novel View Synthesis in High Dynamic Range which has been accepted by ICLR 2026.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published