Install ComfyStream

​

System Prerequisites

Install the necessary software depending on your platform

​

Create Directories to Store Models

1. Choose a location to store AI models and output files from ComfyUI. For example will use:

• Models: ~/models/ComfyUI--models
• Output: ~/models/ComfyUI--output

2. Download the docker image:

docker pull livepeer/comfyui-base:server

​

Run the container

3. Run the container using the paths you created above:

The --download-models and --build-engines flags will download the required models and build TensorRT engines. This process may take some time. You will only need these flags on the first run of the container, or when adding additional models.

​

Access ComfyUI and ComfyStream

The --server flag will start ComfyUI, ComfyStream and UI automatically. You can access ComfyUI and ComfyStream at:

• ComfyUI: http://localhost:8188
• ComfyStream: http://localhost:8888
• UI: https://localhost:3000