| task_categories: | |
| - image-to-video | |
| tags: | |
| - 3d | |
| - gaussian-splatting | |
| - artifact-restoration | |
| # DL3DV-Res Benchmark | |
| [**Project Page**](https://gvclab.github.io/GSFixer/) | [**Paper**](https://huggingface.co/papers/2508.09667) | [**GitHub**](https://github.com/GVCLab/GSFixer) | |
| DL3DV-Res is a benchmark dataset introduced in the paper "GSFixer: Improving 3D Gaussian Splatting with Reference-Guided Video Diffusion Priors". It contains artifact frames rendered using low-quality 3D Gaussian Splatting (3DGS) from sparse views, and is designed for evaluating 3DGS artifact restoration and sparse-view 3D reconstruction. | |
| ## Dataset Usage | |
| The DL3DV-Res benchmark data can be downloaded using the following command, as indicated in the official GitHub repository: | |
| ```bash | |
| python download/download_data_hf.py | |
| ``` | |
| ## Citation | |
| If you find this dataset useful for your research, please consider citing the original paper: | |
| ```bibtex | |
| @article{yin2025gsfixer, | |
| title={GSFixer: Improving 3D Gaussian Splatting with Reference-Guided Video Diffusion Priors}, | |
| author={Yin, Xingyilang and Zhang, Qi and Chang, Jiahao and Feng, Ying and Fan, Qingnan and Yang, Xi and Pun, Chi-Man and Zhang, Huaqi and Cun, Xiaodong}, | |
| journal={arXiv preprint arXiv:2508.09667}, | |
| year={2025} | |
| } | |
| ``` |