Generative Refocusing: Flexible Defocus Control from a Single Image

---

⚡ Quick Start

Follow the steps below to set up the environment and run the inference demo.

1. Installation

Clone the repository:

git clone git@github.com:rayray9999/Genfocus.git
cd Genfocus

Environment setup:

conda create -n Genfocus python=3.12
conda activate Genfocus

Install requirements:

pip install -r requirements.txt

2. Download Weights

You can download the pre-trained models using the following commands. Ensure you are in the Genfocus root directory.

# 1. Download main models to the root directory
wget https://huggingface.co/nycu-cplab/Genfocus-Model/resolve/main/bokehNet.safetensors
wget https://huggingface.co/nycu-cplab/Genfocus-Model/resolve/main/deblurNet.safetensors

# 2. Setup checkpoints directory and download auxiliary model
mkdir -p checkpoints
cd checkpoints
wget https://huggingface.co/nycu-cplab/Genfocus-Model/resolve/main/checkpoints/depth_pro.pt
cd ..

3. Run Gradio Demo

Launch the interactive web interface locally:

Note: This project uses FLUX.1-dev. You must request access and authenticate locally before running the demo.
⚠️ VRAM warning: GPU memory usage can be high. On an NVIDIA A6000, peak usage may reach ~45GB depending on resolution and settings.
Tip (Advanced Settings): If you want faster inference, try lowering num_inference_steps, resizing the input in Advanced Settings.

python demo.py

The demo will be accessible at http://127.0.0.1:7860 in your browser.

---

🧩 ComfyUI Integration

A ComfyUI implementation of Genfocus is available, thanks to Eric Rollei!
Check it out here: 👉 comfyui-refocus

---

🗺️ Roadmap & TODO

We are actively working on improving this project. Current progress:

• Upload Model Weights
• Release HF Demo & Gradio Code (with tiling tricks for high-res images)
• Release Inference Code (Support for adjustable parameters/settings)
• Release Benchmark data
• Release Training Code and Data

---

🔗 Citation

If you find this project useful for your research, please consider citing:

@article{Genfocus2025,
  title={Generative Refocusing: Flexible Defocus Control from a Single Image},
  author={Tuan Mu, Chun-Wei and Huang, Jia-Bin and Liu, Yu-Lun},
  journal={arXiv preprint arXiv:2512.16923},
  year={2025}
}

📧 Contact

For any questions or suggestions, please open an issue or contact me at raytm9999.cs09@nycu.edu.tw.

Star 🌟 this repository if you like it!