This guide assumes you have Python 3.8+ and NVIDIA GPU drivers (CUDA 13.0 or compatible) installed on your system.
1. Navigate to your desired project directory:
Open your terminal and change to the directory where you want to install ComfyUI. If the directory doesn't exist, create it.
1 cd /path/to/your/comfyui_project
2 # If the directory is empty or you want a fresh start, ensure it is empty
3 # If .venv exists, remove it: rm -rf .venv
2. Clone the ComfyUI repository:
This will download all the necessary ComfyUI files into your current directory.
1 git clone https://github.com/comfyanonymous/ComfyUI.git .
3. Create a Python Virtual Environment:
We'll use uv to create an isolated environment for ComfyUI's dependencies.
1 uv venv
4. Install PyTorch with CUDA Support:
This step is crucial for GPU acceleration. We'll use the specific PyTorch build that is compatible with your DGX Spark's CUDA environment (cu129 index).
1 uv pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu129
5. Install ComfyUI's remaining dependencies:
1 uv pip install -r requirements.txt
6. Download the Stable Diffusion Model:
You need a Stable Diffusion model checkpoint. We'll download the v1-5-pruned-emaonly.safetensors model, which is a standard choice.
First, ensure the target directory exists:
1 mkdir -p models/checkpoints
Then, download the model:
1 wget -O models/checkpoints/v1-5-pruned-emaonly.safetensors https://huggingface.co/runwayml/stable-diffusion-v1-5/resolve/main/v1-5-pruned-emaonly.safetensors
7. Launch the ComfyUI Server:
To make ComfyUI accessible over your network (for tools like Open-WebUI), launch it with the --listen flag. You can run it in the foreground to see output or in the background:
* To run in foreground (for debugging/monitoring):
1 .venv/bin/python main.py --listen
(You'll need to press Ctrl+C to stop it.)
* To run in background (recommended for service):
1 nohup .venv/bin/python main.py --listen > comfyui.log 2>&1 &
This command runs ComfyUI in the background, redirects its output to comfyui.log, and detaches it from your terminal. You can check tail -f comfyui.log for logs or use fg to
bring it back to foreground if launched using &.
8. Verify ComfyUI is Running:
Open your web browser and go to http://<YOUR_SPARK_IP>:8188. Replace <YOUR_SPARK_IP> with the actual IP address of your machine. If you see the ComfyUI interface, the server is
running.
9. Test Image Generation:
Inside the ComfyUI web UI, try to generate a simple image using the default workflow. If it successfully generates an image, your Stable Diffusion service is fully operational.
10. Connect to Open-WebUI:
You can now configure Open-WebUI to use this ComfyUI instance by pointing its "API URL" setting to http://<YOUR_SPARK_IP>:8188.
