gokaygokay/Flux-Detailer-LoRA

🎨 Flux Creativity LoRA

This LoRA model is designed to elevate your images by removing the "generic look" and adding a touch of creativity.

🚀 Recommendations

• Use this LoRA with:
  • 0.5 weight and CFG 3.5
  • or 1.0 weight and CFG 2.5

🌟 Example

Prompt: "Beautiful lady in forest, close up shot, extremely detailed, highly detailed"

Settings:

• CFG: 2.5
• LoRA Weight: 1.0

Before

After

Settings:

• CFG: 3.5
• LoRA Weight: 0.5

Before

After

✨ Results

As you can see, this LoRA transforms your images by:

• Enhancing creativity
• Removing generic elements
• Adding unique and eye-catching details

Academic use and citation

This adapter is used in the BBC-PAIR-LoRA evaluation set introduced by Towards Reliable Identification of Diffusion-based Image Manipulations (RADAR), published at NeurIPS 2025.

If this model is useful in your research, please cite both the model and the paper that documents its academic use:

@software{aydogan2024flux_detailer_lora,
  author = {Aydoğan, Gökay},
  title = {Flux Detailer LoRA},
  year = {2024},
  url = {https://huggingface.co/gokaygokay/Flux-Detailer-LoRA},
  note = {Hugging Face model}
}

@article{costanzino2025radar,
  author = {Costanzino, Alex and Bayliss, Woody and Sock, Juil and Gorriz Blanch, Marc and Horak, Danijela and Laptev, Ivan and Torr, Philip and Pizzati, Fabio},
  title = {Towards Reliable Identification of Diffusion-based Image Manipulations},
  journal = {arXiv preprint arXiv:2506.05466},
  year = {2025},
  doi = {10.48550/arXiv.2506.05466}
}

Maintainer: Gökay Aydoğan