grrdhdz-ComfyUI-NVIDIA-Maxine/docs/installation.md

Installation

These nodes are designed to be cloned into a local ComfyUI custom_nodes folder. They use ComfyUI's V3 custom node API.

Requirements

• Windows with Docker Desktop and the WSL2 engine enabled.
• NVIDIA GPU with Docker GPU access working.
• ComfyUI with the V3 custom node API. The package has been developed against ComfyUI 0.21.0.
• Python packages from requirements.txt.
• An NGC Personal API key with access to the NVIDIA Studio Voice NIM image.

Studio Voice is the only supported workflow in this package.

Docker Desktop

Install Docker Desktop for Windows from Docker's official installation guide:

https://docs.docker.com/desktop/setup/install/windows-install/

Use Docker Desktop's WSL2 backend. Docker's GPU support documentation states that GPU support in Docker Desktop is available on Windows with the WSL2 backend:

https://docs.docker.com/desktop/features/gpu/

Start Docker Desktop before running the ComfyUI setup node. The nodes call docker.exe from Windows; users do not need to open a WSL shell.

NVIDIA NGC API Key

Studio Voice uses NVIDIA's local NIM Docker image:

nvcr.io/nim/nvidia/studio-voice:latest

NVIDIA's Studio Voice NIM getting-started page documents the NGC API key and Docker login flow:

https://docs.nvidia.com/nim/maxine/studio-voice/latest/getting-started.html

Before the first pull:

1. Sign in to NVIDIA/NGC.
2. Open the Studio Voice deploy page and accept terms if prompted: https://build.nvidia.com/nvidia/studiovoice/deploy
3. Create an NGC Personal API key: https://org.ngc.nvidia.com/setup/api-keys
4. Include at least the NGC Catalog service for the key.

For Docker login, the username is the literal $oauthtoken; the password is the NGC API key. The ComfyUI setup node handles this for normal users.

Manual Install

From PowerShell:

cd C:\path\to\ComfyUI\custom_nodes
git clone https://github.com/grrdhdz/ComfyUI-NVIDIA-Maxine.git
cd ComfyUI-NVIDIA-Maxine
..\..\python_embeded\python.exe -m pip install -r requirements.txt

For Comfy Desktop or a virtualenv-based install, use the Python executable that launches that ComfyUI installation:

C:\path\to\ComfyUI\.venv\Scripts\python.exe -m pip install -r requirements.txt

Restart ComfyUI after installing dependencies.

ComfyUI Manager

When installed through ComfyUI Manager or cloned manually, the important file is requirements.txt. There is no requirements.py; Python dependencies for ComfyUI custom nodes are normally declared in requirements.txt.

If nodes do not appear after cloning:

1. Confirm the repository folder is directly under custom_nodes.
2. Install requirements.txt with the same Python used by ComfyUI.
3. Restart ComfyUI and check the startup log for import errors.

Docker/NIM Setup

Docker images are not downloaded during Python package installation. The ComfyUI setup node performs Docker login, image pull, container start, and health checks when the user runs it.

For Studio Voice:

1. Add NVIDIA Studio Voice Docker Setup.
2. Paste the NGC API key, or leave the field empty and launch ComfyUI with NGC_API_KEY set in the environment.
3. Run setup_all_transactional.
4. Connect studio_voice_connection to NVIDIA Studio Voice Enhance.

Saved ComfyUI workflows may retain widget values. For shared workflows, do not save API keys inside nodes.

Example Workflow

This repository includes an importable ComfyUI workflow:

workflows/nvidia_studio_voice_enhance.json

Import it with Workflow -> Open or drag it onto the ComfyUI canvas. The API key field is blank by design.