Set up & Run ComfyUI-Copilot on Linux / Ubuntu

ComfyUI Copilot represents a sophisticated, modular graphical user interface (GUI) and backend framework designed for the orchestration of Stable Diffusion models.

This system leverages a graph- and node-based configuration, enabling users to construct and implement intricate image synthesis workflows with minimal direct coding intervention.

System Prerequisites

Before proceeding with installation, ensure that the Linux environment meets the following technical specifications:

Python 3.10 or later – Ensures compatibility with the ComfyUI framework.
Git – Required for repository cloning and version control.
PyTorch – Essential for model computation; installation parameters depend on the hardware architecture.
NVIDIA GPU (recommended) – While the framework can operate on CPUs, an NVIDIA GPU with a minimum of 8GB VRAM significantly enhances computational efficiency.

Installation Protocol

Step 1: System Update

Execute the following command to synchronize package dependencies with the latest available versions:

sudo apt update && sudo apt upgrade -y

Step 2: Installation of Required Dependencies

Deploy Python, pip, and Git using the package manager:

sudo apt install python3 python3-pip git -y

Step 3: Repository Cloning

Retrieve the ComfyUI source code via Git:

git clone https://github.com/comfyanonymous/ComfyUI.git

cd ComfyUI

Step 5: Dependency Installation

Execute the following command to install all necessary Python dependencies:

pip3 install -r requirements.txt

Step 6: PyTorch Installation

For NVIDIA GPUs:

pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu113

For AMD GPUs:

pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/rocm5.2

Step 7: Model Deployment

Download Stable Diffusion model checkpoints (.ckpt or .safetensors) and position them within the models/checkpoints directory.

Step 8: Execution of ComfyUI

Launch ComfyUI by executing:

python3 main.py

Following initiation, the user interface becomes accessible at http://localhost:8188.

Configuration and Advanced Usage

WebUI Setup

Upon launching ComfyUI, navigate to http://localhost:8188 and enable Developer Mode to unlock advanced functionalities.

API Utilization

ComfyUI features an integrated API for automation and external integrations. To enable it:

Access API settings via the WebUI.
Click Save (API Format) to generate a JSON configuration file.

Executing image synthesis via API:

import requests
import json

with open('xyz_template.json') as f:
    data = json.load(f)

response = requests.post('http://localhost:8188/api/run', json=data)
print(response.json())

Ensure ComfyUI is active prior to execution.

Model and Extension Management

Directory Structure

Organize model files within designated subdirectories:

/checkpoints – Primary Stable Diffusion models (e.g., SDXL, Flux)
/vae – Variational autoencoders
/controlnet – ControlNet configurations
Additional folders for LoRA, CLIP, and super-resolution models

Model Installation

Acquire .safetensors models from reputable sources.
Place them in their corresponding directories.
Restart ComfyUI to load new models.

Specialized Configurations

For complex models such as Flux, additional CLIP encoders and VAE settings may be required.

Applied Use Cases

Example 1: Automated Image Synthesis

To programmatically generate images using ComfyUI:

import requests

payload = {
    "model": "stable-diffusion",
    "prompt": "A futuristic cityscape with neon lights",
    "steps": 50,
    "width": 512,
    "height": 512
}

response = requests.post('http://localhost:8188/api/generate', json=payload)
print(response.json())

Example 2: Batch Image Processing

For processing multiple images within a single session:

import requests

prompts = [
    "A fantasy castle in the clouds",
    "A cyberpunk street with rain reflections",
    "A medieval warrior with a sword"
]

for prompt in prompts:
    payload = {"model": "stable-diffusion", "prompt": prompt, "steps": 50, "width": 512, "height": 512}
    response = requests.post('http://localhost:8188/api/generate', json=payload)
    print(response.json())

Diagnostic and Troubleshooting Procedures

Common Issues

Dependency Resolution Errors – Reinstall missing packages from requirements.txt.
GPU Compatibility Conflicts – Verify that GPU drivers align with installed PyTorch versions.
Python Version Mismatch – Ensure Python 3.10 or later is in use.

Conclusion

By adhering to this systematic approach, users can successfully deploy and optimize ComfyUI Copilot on Linux. Proper configuration ensures maximal computational efficiency, allowing for high-performance AI-driven image synthesis.