ComfyUI Easy Install Guide 2025 — Full Setup & First Workflow

What Is ComfyUI?

ComfyUI is a node-based interface for AI image generation using Stable Diffusion and Flux models. Unlike Automatic1111 (which hides most settings behind a form), ComfyUI exposes every step of the image generation process as a visual node you can connect, rearrange, and extend. It's more complex to start, but gives you complete control and makes it the tool of choice for automation workflows.

Why You Want This

• Run image generation locally — no per-image API costs
• Connect to n8n via API for fully automated pipelines
• Use any Stable Diffusion or Flux checkpoint
• Install community nodes to extend what it can do

Installation (Windows — Easy Install Version)

1. Visit the ComfyUI GitHub page (link in the video description). Scroll to Download the Latest Release.
2. Click the Windows download link. Save the ZIP to your drive (e.g. your F: or G: drive — somewhere with plenty of space).
3. Right-click the ZIP → Extract All → leave the destination as-is → Extract.
4. Rename the extracted folder to something short like ComfyUI.
5. Double-click into the folder, then double-click ComfyUI Easy Install to open the installer. Click Allow when Windows asks for permission. This downloads and sets up all dependencies — expect a few minutes.
6. Once complete, press any key to close the installer window. You'll see a new ComfyUI Easy Install subfolder was created.

Running ComfyUI

1. Open the ComfyUI Easy Install folder.
2. Double-click Run Nvidia GPU (or the CPU/AMD equivalent for your hardware).
3. A terminal window opens. ComfyUI launches and prints its local URL — usually 127.0.0.1:8188.
4. Your browser should open automatically, or navigate there manually. You'll see the ComfyUI interface.

First launch takes longer as it sets up the Python environment. Subsequent launches are much faster.

Getting Your First Model

ComfyUI doesn't ship with any models. You need to download a checkpoint:

1. Click the dropdown in the top bar and select Browse Templates.
2. Choose Images and pick a template — the video uses Flux Dev FP8 (a good all-around model).
3. ComfyUI will tell you the checkpoint is missing. Click Download.
4. When prompted to choose a folder, navigate to your ComfyUI folder → models → checkpoints. Click Save.
5. Flux Dev FP8 is ~16GB. Wait for the download to complete.
6. Once done, press R in the browser to refresh — the model will now appear in the checkpoint dropdown.

Understanding the Node System

A default ComfyUI workflow has these nodes connected in order:

• Load Checkpoint — Selects the AI model to use
• CLIP Text Encode (Positive Prompt) — Your image description
• CLIP Text Encode (Negative Prompt) — What to avoid (Flux doesn't use this; other models do)
• KSampler — Controls how the image is generated (steps, CFG, sampler algorithm)
• VAE Decode — Converts the latent space output into a viewable image
• Save Image / Preview Image — Outputs the result

You can rename any node by clicking its title. Right-click on the canvas to add new nodes. Connect nodes by dragging from an output socket (right side of a node) to an input socket (left side of another).

Running Your First Generation

1. Make sure your model is selected in the Load Checkpoint node (it should auto-select once downloaded).
2. Click inside the positive prompt node and type a description.
3. Click the green Run button at the top.
4. Watch the progress bar. The terminal window also shows generation progress.
5. Your image appears in the Save/Preview Image node when done.

What's Next

Once you're comfortable generating images manually, the next step is connecting ComfyUI to n8n so it can generate images automatically as part of a larger workflow. You export your workflow as API JSON (File → Export API) and send it via HTTP POST from n8n.