mmartial/ComfyUI-Nvidia-Docker

ComfyUI for NVIDIA GPU in a Docker container with user configurable uid/gid

What it solves

This project provides a Docker-based deployment for ComfyUI, a Stable Diffusion WebUI. It solves the complexity of setting up the environment, managing NVIDIA GPU drivers, and handling Python dependencies across different hardware generations (from GTX 10xx to RTX 50xx/Blackwell) while ensuring host OS separation and proper user permission management.

How it works

It uses official NVIDIA CUDA containers as a base to ensure optimal GPU performance. The system separates the installation into two main directories: a run folder for the virtual environment and source code, and a basedir folder for user files, models, and custom nodes. It supports both Docker and Podman and allows users to map their own UID/GID to avoid permission issues with files created inside the container.

Who it’s for

Users of ComfyUI who want a portable, isolated installation that is easy to update and compatible with various NVIDIA GPUs, including those on Windows (via WSL2), Linux, and Unraid.

Highlights

  • Hardware-Specific Tags: Provides multiple Ubuntu and CUDA version combinations to support a wide range of GPUs, including specific tags for Blackwell (RTX 50xx) and older GTX 10xx cards.
  • Permission Management: Drops privileges to a regular user and preserves host permissions via custom UID/GID mapping.
  • Flexible Installation: Supports uv for faster package installation and includes a TORCH_LOCK mechanism to prevent breaking updates to PyTorch.
  • Integrated Management: Comes with ComfyUI-Manager pre-installed for easier updates and node management.
  • Customization: Allows command-line overrides via environment variables and the execution of custom bash scripts within the container.

Related

  • Project
    YanWenKun/ComfyUI-Docker

    Docker images for ComfyUI that provide pre-configured environments for NVIDIA, AMD, Intel, and CPU setups to simplify deployment of the node-based AIGC GUI.

  • Project
    Comfy-Org/Comfy-Desktop

    The official desktop application for ComfyUI that provides a self-contained, GPU-ready environment to install, run, and manage multiple independent AI generative workflows without dependency conflicts.

  • Project
    facefusion/facefusion-docker

    A Docker-based deployment for the FaceFusion face manipulation platform, allowing users to easily run the software across various hardware accelerators like CUDA and ROCm.

  • Project
    Comfy-Org/comfy-cli

    comfy‑cli is a Python command‑line tool that installs, launches, and manages ComfyUI (an open‑source generative‑media engine). It handles environment setup, custom‑node installation, model downloads, workflow execution, and can offload jobs to the hosted Comfy Cloud service. Features include fast `uv`‑based dependency resolution, JSON‑structured output for LLM agents, shell completion, and utilities for testing PRs and managing background servers.

  • Project
    YanWenKun/ComfyUI-Windows-Portable

    A ready‑to‑run Windows package that bundles ComfyUI (a node‑based UI for Stable Diffusion) with 40+ custom nodes and 300+ pre‑compiled Python packages, letting users launch AI image/video generation instantly on an NVIDIA GPU.