In this guide, we will show you how to install ComfyUI and use it to create stunning generative art with Stable Diffusion. ComfyUI is a powerful and flexible web UI that lets you create realistic images from text or other images using Stable Diffusion, a state-of-the-art technique for image synthesis. ComfyUI has many advantages over other Stable Diffusion GUIs, such as Automatic1111. For example:

• ComfyUI is more efficient and can run on weaker computers that may not meet the system requirements of Automatic1111.
• ComfyUI allows advanced users to create smooth animations with Stable Video Diffusion, a feature that is not available in Automatic1111.
• ComfyUI has a node-based interface that lets you design and execute complex Stable Diffusion workflows without coding. You can also share your workflows with the community and download other people’s workflows for specific tasks.
• ComfyUI supports various Stable Diffusion models, such as SD1.x, SD2.x, SDXL, and inpainting models. You can also use standalone VAEs, CLIP models, LoRAs, Hypernetworks, ControlNet, T2I-Adapter, unCLIP, GLIGEN, and more.
• ComfyUI is fast, transparent, and works fully offline.

The only downside of ComfyUI is that it can be intimidating and hard to learn for beginners, as you need to understand what each node does and how to use it. However, there are many resources and tutorials available online to help you get started. Before you get started, you should install the prerequisites before continuing. Without it, it won’t work. Please check out this guide below on how to install all the requirements.

---

Step 1: Initiate Download

Start by visiting this GitHub link and scrolling down until you find the “Direct link to download”.
Clicking on it will initiate the download. Alternatively, you can directly download below:

Step 2: Extract the Downloaded File

• Upon completion, you’ll find a file titled
  “ComfyUI_windows_portable_nvidia_cu118_or_cpu”.
• Move this file to your desired installation folder. (This is where your installation files will be located.)
• Extract it using your preferred software (e.g. 7zip)

Step 3: Launch ComfyUI

• Open the “ComfyUI_windows_portable” folder.
• Inside, you’ll find two Launcher options.
  • 1. “run_cpu” for users with less powerful GPUs
  • 2. “run_nvidia_gpu” for those with capable GPUs
  • It should automatically launch the web UI. If not, then to see the GUI:
    You need to search your Command Window for a link like below.
    http://127.0.01:8188 (Yours will probably be different).
  • This will launch the ComfyUI WebUI Instantly. It’s very efficient!

At this point, you’re not done yet, because you’re still missing a very important extension called the ComfyUI Manager, which we will be installing. We also don’t have any models to work with, so we’ll need to load in the Stable Diffusion models as well.

• You’re not done!
  • Close ComfyUI WebUI and the Command window.
  • We will begin installing the ComfyUI Manager.

Step 4: Installing the ComfyUI Manager

The ComfyUI Manager is a tool that helps you manage the custom nodes and workflows in ComfyUI. It allows you to install, update, enable, disable, and delete the nodes and workflows that you need. It also provides a web interface where you can see the information and documentation of all the available nodes and workflows. The ComfyUI Manager is not part of the original ComfyUI software and is developed by a third-party developer named ltdrdata, but it is pretty much a standard tool you should have in the UI so it is part of this installation process.

In this step, I will use Github Desktop to update the file. It is the least technical way to do it.

You can learn how to use Github Desktop here: Prerequisites Installation Guide

To clone and install the ComfyUI Manager repository, follow these steps:

• Open Github Desktop:
  • Go to File > Clone Repository > URL
• Copy and paste this URL in the Repository URL space:
  • https://github.com/ltdrdata/ComfyUI-Manager
• In Local Path:
  • Go to your ComfyUI installation folder > ComfyUI > custom_nodes folder
• Click Clone and wait for the process to finish.
• Launch Comfy UI again.
  • 1. “run_cpu” for users with less powerful GPUs
  • 2. “run_nvidia_gpu” for those with capable GPUs
  • If the WebUI doesn’t automatically launch:
    You need to search your Command Window for a link like below.
    http://127.0.01:8188 (Yours will probably be different).
  • Copy and Paste that URL to your Internet Browser.
  • This will launch the ComfyUI WebUI Instantly. It is very efficient!

You have successfully cloned the ComfyUI Manager repository. You can now use it to install and update custom nodes and workflows in ComfyUI.

With the ComfyUI Manager, you can do the following:

• Install custom nodes that you need for your stable diffusion projects
• Install missing custom nodes that are required by other user’s workflows
• Update ComfyUI to the latest version and get new features and bug fixes
• Install models that are compatible with different versions of stable diffusion

The ComfyUI Manager is a useful tool that makes your work easier and faster, most importantly, it allows you to install nodes and workflows from the community. I will explain more about it in a future blog post. Stay tuned!

Step 5: Set Up Models

Option 1- Using models from A1111 folder.

If you’re an Automatic1111 user with various models in the Stable Diffusion Models folder:

Option 2- Download your models from Civitai.com

For this option, Do not change the “extra_model_paths.yaml.example”. Leave it as is. We will install both Stable Diffusion 1.5 and SDXL 1.0 for this guide, but you can download whatever you want. SD 1.5 is the popular option, while SDXL 1.0 allows you to generate higher resolution. I recommend playing with both.

• Download the Stable Diffusion SDXL 1.0 Base Models:

  SDXL Base SDXL Base (vae) SDXL Offset LoRa

• Download the Stable Diffusion SDXL 1.0 Refiner Models:

  SDXL Refiner SDXL Refiner (vae)

• Download the Stable Diffusion SD 1.5 Models:

  SD 1.5 pruned emaonly SD 1.5 pruned

• Download the Stable Video Diffusion Models:

  SVD SVD Image Decoder

• Download community fine-tuned models

  

  www.civitai.com

• Download community fine-tuned models at Civitai.com

  Navigate to your ComfyUI folder

  Then “ComfyUI > models > checkpoints”, and place your models there.

• Launch ComfyUI

  • 1. “run_cpu” for users with less powerful GPUs
  • 2. “run_nvidia_gpu” for those with capable GPUs
  • It should automatically launch the web UI.
  • If the WebUI doesn’t automatically launch:
    You need to search your Command Window for a link like below.
    http://127.0.01:8188 (Yours will probably be different).
  • Copy and Paste that URL to your Internet Browser.
  • This will launch the ComfyUI WebUI Instantly.

• Done. Enjoy ComfyUI!

---

Source: SDXL 1.0, SDXL Refiner, SD 1.5, Stable Video Diffusion

---

Final 6: Updating ComfyUI (Optional Step)

ComfyUI also provides an easy update mechanism.

1. To update, simply navigate to the update folder:
   ComfyUK > update > update_comfyui
2. Run the “update_comfyui” batch file, and your ComfyUI will be updated!
3. Update using ComfyUI Manager
   • Alternatively, you should have the ComfyUI Manager already installed.
   • Launch the Manager and click on Update ComfyUI, or Update All to update everything.
     (Update everything updates all the nodes and extensions as well)

ComfyUI Hot Keys

For more information on ComfyUI, take a look at their Github.

How to Install Stable Diffusion:

• How to Install Stable Diffusion Web UI (Complete Installation)
• Prerequisites Installation Guide to Stable Diffusion’s Web UI’s
• How to Install Stable Diffusion SDXL 1.0: A Step-by-Step Guide
• How to Install Automatic1111 Web UI for Stable Diffusion
• How to Install ComfyUI: A Streamlined Process for Enhanced Workflow
• How to Install InvokeAI 3.0: A Cleaner Web UI for Stable Diffusion