Although Fooocus aims to provide a simple and intuitive interface, it can't ignore the advancements in Stable Diffusion and the new uses of generative AI.The application might have seemed neglected by its creator, but a collaborator, mashb1t, has taken over in recent months and is handling the updates. The recently released version 2.5.0 owes much to mashb1t's efforts—let's take a closer look at what it has to offer.

How to update Fooocus ?

Typically, Fooocus updates automatically when you run it using the run.bat file on Windows or the entry_with_update.py script on Mac. If the automatic update doesn't happen, you'll need to update manually using one of the following commands:

• For Windows: ..\python_embeded\python.exe -m pip install -r .\requirements_versions.txt
• For Mac: python.exe -m pip install -r requirements_versions.txt

These commands will install the necessary dependencies as specified in the requirements_versions.txt file.

What's New in Fooocus 2.5.0 ?

Enhance: Correct and Improve

The new "Enhance" feature in Fooocus 2.5.0 allows users to enlarge, improve, or correct images or specific parts of them. Similar to ADetailer, Enhance offers greater flexibility and better object detection. It is particularly effective for refining and enhancing faces and intricate details in an image:

Enhance also enables targeted and automatic inpainting to modify specific details, such as changing the color of a dress or correcting hand details:

As with many features in Fooocus, Enhance leverages a combination of specialized models to achieve the best possible results. The primary models used are SAM (Segment Anything) and GroundingDINO, though Fooocus also supports other models.

New Models and Presets

Some presets in Fooocus have been updated to utilize the latest versions of the models:

• The Anime preset now uses AnimaPencilXL_v500.
• The Realistic preset has been updated to use LoRA SDXL_FILM_PHOTOGRAPHY_STYLE_V1.
• A new preset, pony_v6, has been introduced, featuring the popular ponyDiffusionV6XL. This preset includes a "Fooocus Pony" style, which automatically applies essential tags for this model, such as "score_9," "score_8_up," and "score_7_up."

Other Notable Updates

• A new checkbox to enable inpainting masks is now available in the Inpainting or Outpainting tab.
• Image editing features have been added for uploaded inpainting masks, allowing for more precise control over edits.
• The inclusion of VAE (Variational Autoencoder) has been added as a configurable option within presets, giving users more flexibility in adjusting image generation settings.

🔗 For more details, please refer to the release notes for versions 2.5.0 on Fooocus's GitHub page.

ℹ️ This article is a translation from the Fooocus 2.5 announcement on the French Stable Diffusion Blog.