How to install ComfyUI

If you would like to install Videcool on your system, you also need to setup ComfUI. ComfyUI isn’t just another front-end, it’s the actual AI-model execution engine that Videcool relies on to generate images and videos using diffusion pipelines. Installing ComfyUI gives Videcool the runtime backbone it needs; without it, Videcool has nowhere to plug in its nodes and no way to fire the GPU kernels that turn your text prompts into frames. In this guide you can find the installation steps needed to setup ComfuUI on your computer.

Installation steps

Copy-paste ready steps

Install CUDA

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/x86_64/cuda-ubuntu2404.pin
sudo mv cuda-ubuntu2404.pin /etc/apt/preferences.d/cuda-repository-pin-600
wget https://developer.download.nvidia.com/compute/cuda/12.8.0/local_installers/cuda-repo-ubuntu2404-12-8-local_12.8.0-570.86.10-1_amd64.deb
sudo dpkg -i cuda-repo-ubuntu2404-12-8-local_12.8.0-570.86.10-1_amd64.deb
sudo cp /var/cuda-repo-ubuntu2404-12-8-local/cuda-*-keyring.gpg /usr/share/keyrings/
sudo apt-get update
sudo apt-get -y install cuda-toolkit-12-8
sudo apt-get -y install nvidia-gds

export PATH=${PATH}:/usr/local/cuda-12.8/bin
export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/usr/local/cuda-12.8/lib64

Install Conda

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash ~/Miniconda3-latest-Linux-x86_64.sh
source ~/.bashrc

Install ComfyUI

conda create -n ComfyUI python=3.12.11
conda activate ComfyUI

pip install comfy-cli

comfy --install-completion

comfy install

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

Start ComfyUI

comfy launch -- --listen 0.0.0.0 --port 8000

Create ComfyUI service

cd /etc/systemd/system

sudo nano comfyui.service

[Unit]
Description=comfyui
After=network.target

[Service]
User=user
WorkingDirectory=/home/user/comfy/ComfyUI
ExecStart=bash -c 'source /home/user/miniconda3/bin/activate ComfyUI; \
comfy launch -- --listen 0.0.0.0 --port 8000 --enable-cors-header "*"'
Restart=always
TimeoutStartSec=700

[Install]
WantedBy=multi-user.target

sudo systemctl daemon-reload

sudo systemctl start comfyui.service

Step 1: Enable the NVIDIA driver

To generate Images and Videos with AI you need an NVIDIA GPU. Linux needs the NVIDIA drivers enabled to make it possible to work with the GPU.

Step 2: Install NVIDIA CUDA toolkit

Once the NVIDIA driver is enabled, you need to Install the NVIDIA toolkit. This software package released by NVIDIA provides functionality for AI model execution.

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/x86_64/cuda-ubuntu2404.pin
sudo mv cuda-ubuntu2404.pin /etc/apt/preferences.d/cuda-repository-pin-600
wget https://developer.download.nvidia.com/compute/cuda/12.8.0/local_installers/cuda-repo-ubuntu2404-12-8-local_12.8.0-570.86.10-1_amd64.deb
sudo dpkg -i cuda-repo-ubuntu2404-12-8-local_12.8.0-570.86.10-1_amd64.deb
sudo cp /var/cuda-repo-ubuntu2404-12-8-local/cuda-*-keyring.gpg /usr/share/keyrings/
sudo apt-get update
sudo apt-get -y install cuda-toolkit-12-8
sudo apt-get -y install nvidia-gds

export PATH=${PATH}:/usr/local/cuda-12.8/bin
export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/usr/local/cuda-12.8/lib64

Step 3: Install Conda environment

The Conda framework is an environment virtualization system for Python. It makes it possible to run different python programs without version conflicts.

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash ~/Miniconda3-latest-Linux-x86_64.sh
source ~/.bashrc

Step 4: Install ComfyUI

Once all the prerequisites are installed, you are ready to setup ComfyUI on your system.

conda create -n ComfyUI python=3.12.11
conda activate ComfyUI

pip install comfy-cli

comfy --install-completion

comfy install

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128
H2>How to start ComfyUI after installation

To start ComfyUI after installation, run the comfy-cli launch command that brings up the server and listens on all interfaces on the chosen port. Use the following example from the guide: comfy launch -- --listen 0.0.0.0 --port 8000 to start the ComfyUI backend. Once this command is executed, you can connect to the ComfyUI web interface from your browser using the configured host and port.

comfy launch -- --listen 0.0.0.0 --port 8000

How to open the ComfyUI GUI in the browser

To open the ComfyUI GUI in the browser, first make sure the ComfyUI backend is already running and listening on a network interface and port (for example 0.0.0.0:8000). Then, on the same machine or another machine on the network, open a web browser and enter the URL in the address bar in the form http://:, such as http://localhost:8000 or http://192.168.1.10:8000. After pressing Enter, the ComfyUI graphical web interface will load, allowing you to build workflows, configure nodes, and start image or video generation jobs.

How to create a services entry to start comfyui automatically

To create a services entry to start ComfyUI automatically, you need to define a systemd unit file that activates your Conda environment and launches ComfyUI with the desired parameters at boot. In /etc/systemd/system/comfyui.service, set the user, working directory, and an ExecStart command that sources Miniconda, activates the ComfyUI environment, and runs comfy launch -- --listen 0.0.0.0 --port 8000 --enable-cors-header "*" with automatic restart. After saving the file, reload systemd with sudo systemctl daemon-reload and start (and optionally enable) the service using sudo systemctl start comfyui.service so ComfyUI comes up automatically after each system boot.

cd /etc/systemd/system

sudo nano comfyui.service

[Unit]
Description=comfyui
After=network.target

[Service]
User=user
WorkingDirectory=/home/user/comfy/ComfyUI
ExecStart=bash -c 'source /home/user/miniconda3/bin/activate ComfyUI; \
comfy launch -- --listen 0.0.0.0 --port 8000 --enable-cors-header "*"'
Restart=always
TimeoutStartSec=700

[Install]
WantedBy=multi-user.target

sudo systemctl daemon-reload

sudo systemctl start comfyui.service

Summary

To run Videcool’s AI image- and video-generation pipelines you must first install ComfyUI. ComfyUI is the underlying execution engine that turns text prompts into GPU-rendered frames. The process begins by enabling NVIDIA drivers on Ubuntu, then adding the CUDA toolkit and cuDNN libraries that supply the GPU kernels. Next, Miniconda is installed to isolate Python dependencies, after which a fresh Conda environment (Python) is created. Finally, ComfyUI itself is pulled in via the comfy-cli package, PyTorch with CUDA support is layered on top, and the environment is ready for Videcool to plug in its diffusion nodes.


More information