Snoopiam/Fooocus — reverse-engineered prompt

Reverse engineered prompt

Build me a simple local image generator called Fooocus that feels like the easy online tools, but runs offline on my own machine. I want a browser based interface where I can type a short or long prompt and get good looking images without needing to fiddle with lots of settings.

It should handle text to image, upload an image to make subtle or strong variations, upscale images, inpaint and outpaint in different directions, use an image as a prompt, swap faces, and describe an uploaded image. I also want styles, aspect ratios, negative prompts, multiple prompt lines, prompt weighting, and a few easy presets like default, anime, and realistic.

Please make the first launch smooth, with any needed models downloading automatically when they are missing. Keep it usable on a modest Nvidia GPU if possible, and make startup very simple on Windows, ideally just unzip and run. A local web UI is perfect. Look up current docs online if you need to.

Want more depth? Deep Reverse