Haofan Wang* · Peng Xing · Hao Ai · Renyuan Huang · Qixun Wang · Xu Bai
InstantX Team
*corresponding authors
InstantStyle-Plus is a pre-experimental project on the top of InstantStyle and is designed to improve content preservation capabilities in style transfer. We decompose this task into three subtasks: style injection, spatial structure preservation, and semantic content preservation.
- [2024/07/01] 🔥 Code and Techincal Report released.
Our work requires pre-trained checkpoints from InstantStyle, Tile-ControlNet, MistoLine and CSD.
# download adapters
huggingface-cli download --resume-download h94/IP-Adapter --local-dir checkpoints/IP-Adapter
# download ControlNets
huggingface-cli download --resume-download TheMistoAI/MistoLine --local-dir checkpoints/MistoLine
huggingface-cli download --resume-download xinsir/controlnet-tile-sdxl-1.0 --local-dir checkpoints/controlnet-tile-sdxl-1.0
# follow https://github.com/haofanwang/CSD_Score?tab=readme-ov-file#download to download CSD models
git clone https://github.com/haofanwang/CSD_Score
Set HF_ENDPOINT in case you cannot access HuggingFace.
python infer_style.py
Users are granted the freedom to create images using this tool, but they are obligated to comply with local laws and utilize it responsibly. The developers will not assume any responsibility for potential misuse by users.
InstantStyle-Plus is developed by InstantX Team. Our work is built on ReNoise-Inversion and CSD.
If you find InstantStyle-Plus useful for your research and applications, please cite us using following BibTeX:
@article{wang2024instantstyle,
title={InstantStyle-Plus: Style Transfer with Content-Preserving in Text-to-Image Generation},
author={Wang, Haofan and Xing, Peng and Huang, Renyuan and Ai, Hao and Wang, Qixun and Bai, Xu},
journal={arXiv preprint arXiv:2407.00788},
year={2024}
}
@article{wang2024instantstyle,
title={Instantstyle: Free lunch towards style-preserving in text-to-image generation},
author={Wang, Haofan and Wang, Qixun and Bai, Xu and Qin, Zekui and Chen, Anthony},
journal={arXiv preprint arXiv:2404.02733},
year={2024}
}For any question, feel free to contact us via [email protected].