Fooocus

Fooocus is an image generating framework that is based on Stable Diffusion. The goal of Fooocus is to combine the best aspects of Stable Diffusion and Midjourney. It’s designed to reduce the complexity of other Stable Diffusion interfaces, is easy to use and generates high-quality images out of the box.

Here are some key features of Fooocus:

1. Ease of Use: Fooocus is designed to be user-friendly. It only requires a single prompt for the image generation process.

2. Quality Images: Fooocus has optimized the Stable Diffusion pipeline to deliver excellent images. It uses a set of default settings that are optimized to give the best results when using SDXL models.

3. Prompt Expansion: Fooocus expands your prompt with a GPT based prompt engine. This means you don’t need to write long and complicated prompts.

Input Image Types in Fooocus

These are the images you will be using as inputs in Fooocus. You can import various types of images such as an image prompt, Face image, Pyracanny, and CPDS depending on the use-case or desired final image output.

Image prompt

This is a general type of input where an image acts as a reference guide for the AI. It guides the image generation process in the direction of the provided image.

Face Image

This is a specific kind of image prompt where the provided image focuses on a face. It’s useful when you want to specifically modify or generate an image centered around a face.

Pyracanny

Pyracanny is a pyramid-based Canny edge control and it excels at detecting edges in high-resolution images. Regular Canny edge detection might miss some details, so PyraCanny tackles this by analyzing the image at various resolutions to create a more comprehensive edge map. This edge map then influences the final image.

CPDS

Contrast Preserving Decolorization (CPDS) is another custom control method used in Fooocus. It works in a way that’s comparable to ControlNet depth conditioning. CPDS helps in understand the depth information within the image, which can be useful for creating a more realistic or nuanced final image output.