Focus is an open source AI image generation software based on Stable Diffusion, which is popular for its simplicity and efficient image generation capabilities. The following steps will help you complete the installation and preliminary setup.
Before you begin, make sure your system meets the following minimum requirements:
Operating system : Windows 10 or 11
Graphics card : NVIDIA GPU (minimum 4GB VRAM, 6GB or more recommended for better performance); AMD GPU can also run, but support is still in the beta stage and may have low performance.
System memory : At least 8GB RAM
Storage space : at least 10GB of available space (including download of model files)
Visit Focus's official GitHub page: https://github.com/lllyasviel/Fooocus .
Find the "Download" or "Releases" section on the home page and download the latest zip package for Windows (usually in .zip or .7z format).
After the download is complete, save the file to the directory you want to install (such as desktop or D drive).
If your system does not have an unzip tool (such as 7-Zip), please download and install 7-Zip from https://www.7-zip.org/ first.
Right-click on the downloaded Fooocus compressed file and select "7-Zip" > "Extract to Here" or "Extract to [Foooocus Folder Name]".
After decompressing, you will see a folder containing multiple files, such as run.bat , run_anime.bat , etc.
Enter the unzipped folder.
Double-click the run.bat file to start the Fooocus basic version.
First run : The program will automatically download the Stable Diffusion model file (approximately 6-7GB), which may take some time depending on your network speed. Please keep the network connection stable.
After the download is completed, Fooocus will automatically open a web interface in the default browser (usually http://127.0.0.1:7865 ).
If you want to try a specific style, you can run:
run_anime.bat (animation style)
run_realistic.bat (realistic style)
These versions will download the corresponding preset model.
Prompt word input : Enter a prompt word describing the image in the text box at the top of the web interface (for example, "a beautiful sunset over the ocean").
Generate image : Click the "Generate" button and Fooocus will generate an image based on the prompt word. The generation time depends on your hardware performance and usually takes seconds to minutes.
Output location : The generated image is saved in the outputs subfolder under the Focus folder by default.
Check the "Advanced" option in the interface to expand more settings:
Resolution : Adjust the image size (default 512x512, recommended to adjust according to the video memory).
Style template : Select a preset style (such as "cinematic-default" or "SAI Photographic").
Model selection : Add a custom SDXL model in the Focus/models/checkpoints folder and select in the interface.
If you need to adjust the default configuration, you can edit the config.txt file (located in the root directory of Fooocus).
NVIDIA Users : Make sure that the latest NVIDIA drivers and CUDA support are installed.
AMD users : Additional configuration is required (refer to AMD installation instructions on GitHub), and performance may not be as good as NVIDIA.
If the generation speed is slow, try to reduce the resolution or reduce the number of generation steps (Steps).
The DLL file is missing when starting :
It may be that Microsoft Visual C++ Redistributable is missing. Go to Microsoft's official website to download and install the latest version.
Model download failed :
Check the network connection, or manually download the model file and put it into the Focus/models/checkpoints folder.
Generate image stuck :
Make sure that the video memory is sufficient and close other programs that occupy GPU.
Update Focus : Check the GitHub page regularly to get the latest version, download and replace the old files.
Community Support : If you encounter any problems, you can seek help in GitHub Issues or related forums such as Reddit's r/StableDiffusion.
Through the above steps, you should be able to successfully install and set up Fooocus on Windows and start enjoying the fun of AI image generation!