Fooocus

Offline Stable Diffusion interface that hides the complicated settings

AI Image GenerationFree planOverseasβ˜…β˜…β˜…β˜…β˜† 4.0

What is Fooocus?

Fooocus is a free, open-source interface for running Stable Diffusion on your own computer, built by the author of ControlNet. It deliberately strips away most of the knobs: no sampler choice, no CFG ladder, no model mixing screen. You write a prompt, optionally pick a style, and press generate. Underneath it still does sensible things automatically - prompt expansion, quality presets, refiner models and resolutions that suit the loaded checkpoint - and it supports local inpainting, outpainting and image prompting. Installation is scripted for Windows and Linux, and nothing is uploaded anywhere.

Last updated: 2026-09-20. This site only provides an index; for exact features, pricing, and licensing, see the official website.

Key features

  • Local Stable Diffusion generation with no account or cloud upload
  • Automatic prompt expansion and quality presets behind a simple UI
  • Built-in inpainting, outpainting and image prompt support
  • Style templates that adjust the model and settings together
  • One-script install for Windows and Linux with bundled models
  • LoRA and checkpoint support through the standard models folder

Pros & cons

Strengths

  • Simple enough for a first-time diffusion user
  • Free and offline: no credits, no queue, no upload
  • Automatic quality settings reduce obviously bad outputs

Watch out for

  • Requires a GPU with adequate video memory to be pleasant
  • Less control than a node editor when you want precise pipelines
  • Model downloads are large and updates need manual work

Best for & use cases

offline image generation, privacy-conscious users, beginners and local experimentation

If you're comparing similar products, check the alternatives below, or browse all tools in the AI Image Generation category.

FAQ

What hardware does Fooocus need?

An Nvidia GPU with a reasonable amount of video memory gives the smoothest experience, and the project publishes notes on low-memory modes. CPU-only generation works but is slow enough to be frustrating for iteration.

Do I need to know Stable Diffusion settings?

No, that is the design goal. Fooocus chooses sampler, steps and refiner behaviour automatically. Advanced users can still reach checkpoints, LoRAs and inpainting when they need them.

Is anything uploaded?

No. Generation happens on your machine and prompts stay local. The only network traffic is downloading the application and model weights in the first place.