ODS V3 Pre-Release: Public testing and refinement ahead of the official V3 launch. Turn your PC, Mac, or Linux box into a private AI server.
Fooocus
Local SDXL image generation using the official ghcr.io/lllyasviel/fooocus image pinned by digest. Replaces the former RunPod reference recipe. Ships disabled as a first-party GPU service; uploaded recipes do not gain GPU privileges.
Requirements
- GPU: NVIDIA (ODS feature requests 8 GB VRAM), compatible CUDA 12.4 driver and Docker NVIDIA GPU runtime.
- OS: Linux x86_64 or Windows with Docker Desktop WSL2 NVIDIA support. This Docker recipe does not support AMD, CPU inference or macOS Metal.
- Storage/network: Several GB for the image and first model download. Uses a named volume to avoid Windows bind-mount performance issues.
Enable / Disable
ods enable fooocus
ods disable fooocus
Your data is preserved when disabling. To re-enable later: ods enable fooocus
Access
- URL:
http://localhost:7865
First-Time Setup
- Enable the service:
ods enable fooocus - Open
http://localhost:7865 - Start generating images with natural language prompts
First startup downloads model files. The initial health window is 30 minutes; inspect download progress before retrying on slower connections. Later starts still require model loading, so readiness is determined by the HTTP health check.
FOOOCUS_PORT changes the default host port. The default binding is loopback; public access and authentication are not configured by this recipe.
The fooocus-data named volume preserves configuration, models and outputs. Outputs retain upstream's /content/app/outputs path, linked to the volume by its entrypoint so history works. Disable/re-enable preserves the volume.
Provenance and validation
- Official Docker instructions
- Official Compose definition
upstream.jsonrecords the independently verified image digest and architecture. The published image predates the documentation revision; they are not claimed to be the same build.- Code license: GPL-3.0. Downloaded checkpoints have separate licenses.
- Static checks do not prove generation. GPU startup and inference remain unverified; no image or model was downloaded for this change.