lllyasviel/Fooocus ? reverse-engineered prompt

Build me an app for generating images from text prompts, kind of like a free, offline version of Midjourney that I can run on my own computer. The main goal is simplicity, so I should be able to just type in a prompt and get a beautiful image without having to mess with a bunch of complicated settings. The backend should automatically improve my prompts to get better results.

The interface should be a simple web UI using Gradio. Besides basic text-to-image, it needs features like taking an input image and letting me upscale it, create variations, or do inpainting and outpainting. It should also support image prompts and have advanced options for selecting different styles (like anime or realistic), aspect ratios, and maybe even a faceswap tool.

Please build this in Python using the Stable Diffusion XL model. The whole point is to let the user focus on prompting and generating, not on technical tinkering.