Stable Diffusion web UI
uramaki.automatic1111
A Docker Compose file for Stable Diffusion web UI alone, generated from the catalog and checked by the safety inspector. No account needed.
Web interface for Stable Diffusion image generation, covering text-to-image, image-to-image, inpainting, upscaling and model switching, with a large third-party extension ecosystem.
This interface exposes nearly every parameter of a diffusion pipeline — sampler, step count, guidance scale, seed, resolution, denoising strength — as form controls, alongside inpainting and outpainting, batch generation, prompt weighting and attention editing. Checkpoints, LoRA adapters, embeddings and upscalers are dropped into directories and selected from menus, so switching between fine-tuned models does not mean reconfiguring anything. An extension system adds controls such as pose and depth conditioning, and the same features are reachable over an API for scripted use. A first deployment places one checkpoint in the models directory and generates an image from a prompt.
You know it worked when
- The interface loads and lists at least one checkpoint in the model selector.
- A text prompt produces an image.
- Image-to-image and inpainting accept an uploaded source and return a modified result.
- Switching to a second checkpoint completes without restarting the container.
- Generated images are written to the mounted output directory.
Known sharp edges
- It requires a GPU with matching drivers and container runtime configuration; CPU generation technically works but takes minutes per image.
- No model ships with the image, and the interface starts and then fails at generation time when the models directory is empty.
- Extensions are arbitrary code fetched from third-party repositories and executed inside the container with full access to it.
- Model checkpoints carry their own licences, and several widely circulated ones restrict commercial output regardless of the software licence.