This project provides Docker images for running ComfyUI, a popular node-based UI for Stable Diffusion, in cloud and local GPU environments. It targets users who need a pre-configured, easily deployable ComfyUI setup, especially on GPU rental platforms like Vast.ai, simplifying the process of getting started with advanced AI art generation.

How It Works

The images leverage a common AI-Dock base image, designed for cloud-first deployment, which includes authentication and user experience enhancements. ComfyUI itself is run as a service within the container, with options to customize startup arguments and ports. The images are built for NVIDIA CUDA, AMD ROCm, and CPU, with version tags indicating CUDA/ROCm versions and Ubuntu base. Model and third-party configurations are not bundled, requiring a separate provisioning script for setup.

Quick Start & Requirements

• Install/Run: Pull and run a Docker image from GHCR or DockerHub (e.g., docker run --gpus all -p 8188:8188 ghcr.io/ai-dock/comfyui:latest-cuda).
• Prerequisites: NVIDIA GPU with CUDA or AMD GPU with ROCm recommended for performance. Docker installed.
• Configuration: Use environment variables like HF_TOKEN, CIVITAI_TOKEN for gated models, and COFYUI_ARGS for startup flags. Provisioning scripts are needed for model setup.
• Documentation: AI-Dock Base Wiki

Highlighted Details

• Supports NVIDIA CUDA, AMD ROCm, and CPU platforms.
• Environment variables for auto-update, authentication (HuggingFace, Civitai), and ComfyUI startup arguments.
• Includes a ComfyUI API wrapper service.
• Password protection enabled by default for all services.

Maintenance & Community

• Maintained by @robballantyne.
• Links to pre-configured templates for Vast.ai are provided.

Licensing & Compatibility

• The README does not explicitly state the license for the Docker images or the AI-Dock base. ComfyUI itself is typically under an MIT license. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The images do not bundle models or third-party configurations, requiring users to manage these separately via provisioning scripts. The ComfyUI API wrapper is noted as a work-in-progress.