Aesthetic-4K Dataset

We introduce Aesthetic-4K, a high-quality dataset for ultra-high-resolution image generation, featuring carefully selected images and captions generated by GPT-4o. Additionally, we have meticulously filtered out low-quality images through manual inspection, excluding those with motion blur, focus issues, or mismatched text prompts.

For more details, please refer to our paper:

• Diffusion-4K: Ultra-High-Resolution Image Synthesis with Latent Diffusion Models (CVPR 2025)
• Ultra-High-Resolution Image Synthesis: Data, Method and Evaluation
• Source code is available at https://github.com/zhang0jhon/diffusion-4k.

Citation

If you find our paper or dataset is helpful in your research or applications, generously cite with:

@inproceedings{zhang2025diffusion4k,
 title={Diffusion-4K: Ultra-High-Resolution Image Synthesis with Latent Diffusion Models},
 author={Zhang, Jinjin and Huang, Qiuyu and Liu, Junjie and Guo, Xiefan and Huang, Di},
 year={2025},
 booktitle={IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
}

@misc{zhang2025ultrahighresolutionimagesynthesis,
 title={Ultra-High-Resolution Image Synthesis: Data, Method and Evaluation},
 author={Zhang, Jinjin and Huang, Qiuyu and Liu, Junjie and Guo, Xiefan and Huang, Di},
 year={2025},
 note={arXiv:2506.01331},
}