ComfyUI Basics: Nodes, Workflows & Custom Nodes
Understand the core concepts of ComfyUI – the node-based interface for AI image generation.
Table of Contents
01What Is ComfyUI?
ComfyUI is a powerful, node-based interface for Stable Diffusion and other AI models. Unlike simple web UIs with text fields and buttons, ComfyUI works with a visual graph where individual processing steps are represented as 'nodes' and connected to each other. This gives you maximum control over every aspect of the generation.
02The Node Concept
Each node in ComfyUI has a specific task: One node loads the model, another processes the prompt, another performs the sampling, and so on. Nodes are linked through connections. Outputs of one node serve as inputs for the next. The result is a visual pipeline that represents the entire generation process.
03Important Standard Nodes
The fundamental nodes you should know:
- Load Checkpoint: Loads an AI model (e.g., SDXL, Flux) into memory
- CLIP Text Encode: Converts your text prompt into a form understandable by the model
- KSampler: The core of generation – performs the diffusion steps
- VAE Decode: Converts the latent image into a visible image
- Save Image: Saves the generated image to disk
- Load Image: Loads an existing image for img2img or ControlNet
- ControlNet Apply: Applies ControlNet guidance to the generation
04Custom Nodes
The strength of ComfyUI lies in its extensibility. Thousands of community-developed custom nodes extend the functionality: from video generation to face detection to complex image processing pipelines. The ComfyUI Manager greatly simplifies the installation and management of custom nodes.
05Sharing and Loading Workflows
Workflows can be exported and shared as JSON files. To load a workflow, simply drag the JSON file onto the ComfyUI canvas or use the Load function. ComfyVault offers a curated collection of tested workflows that you can download and use directly.
06ComfyUI vs. Automatic1111 / Forge
Compared to A1111 (Automatic1111) or Forge, ComfyUI offers more flexibility and control. A1111 is simpler for beginners with its classic UI but more limited for complex pipelines. ComfyUI is increasingly becoming the standard for advanced users, as it often supports new models and techniques first. The learning curve is steeper, but the possibilities are virtually unlimited.