Search
English
Contents
  1. ComfyUI Tutorial
Getting started
  1. What is ComfyUI
  2. Install ComfyUI
  3. Interface tour
  4. Model downloads and folders
  5. First image (text to image)
Core techniques
  1. KSampler parameters
  2. Prompting techniques
  3. Image to image (img2img)
  4. Inpainting and outpainting
  5. LoRA
  6. ControlNet
  7. Image upscaling
Advanced Topics
  1. Custom Nodes and Manager
  2. Flux and other new-generation models
  3. AI video generation
  4. Automate with the API
  5. FAQ and troubleshooting

ComfyUI Tutorial: Generate AI Images with Node-Based Workflows

ComfyUI is an open-source AI image generation tool that builds workflows from nodes. You can run image models such as Stable Diffusion, SDXL, Flux, and Qwen-Image, as well as video models such as Wan, on your own computer. It’s free with no usage limits, and you can see and adjust every step of the generation process.

[Load model] ─▶ [Enter prompt] ─▶ [Sampler KSampler] ─▶ [Decode VAE] ─▶ [Save image]

Compared with web-based AI image services, ComfyUI has a steeper learning curve, but gives you:

  • Full control: Adjust every step and every parameter
  • Reproducibility: Save and share workflows so others can load and run the same process
  • Fast support for new models: New open-source models are usually available in ComfyUI right away
  • Free and private: Run it locally without a subscription, and your images are never uploaded

What you'll learn

  • Install ComfyUI (Desktop, the portable Windows version, and manual installation)
  • Navigate the interface and understand nodes and connections
  • Generate images from text and images, and use inpainting
  • Configure KSampler parameters and write prompts
  • Use LoRA and ControlNet, and upscale images
  • Install custom nodes, use new-generation models such as Flux, and generate AI video
  • Generate images automatically from code through the API

Hardware requirements

  • NVIDIA GPU: Recommended. 8 GB of VRAM is enough to run SD 1.5 / SDXL smoothly; 12 GB or more is recommended for large models such as Flux and video generation.
  • Apple Silicon Mac: Supported, but slower than a comparable NVIDIA GPU.
  • AMD / Intel GPU: Partial support.
  • Disk space: Model files are large, so set aside at least 50 GB.

Start with What is ComfyUI?.

Getting started

  1. What is ComfyUI

    Learn what ComfyUI is, how node-based workflows work, how it compares with Stable Diffusion WebUI (AUTOMATIC1111 and Forge), and how diffusion models generate images.

  2. Install ComfyUI

    Install ComfyUI with Desktop, the Windows portable build, or Git and Python. Covers GPU options, launch arguments, and updates.

  3. Interface tour

    Learn the ComfyUI interface: canvas controls, adding and searching for nodes, connections and data type colors, the execution queue, opening and saving workflows and templates, bypassing and muting nodes, and the most useful shortcuts.

  4. Model downloads and folders

    Learn where to put ComfyUI models, where to download them, which formats to choose, how model size affects VRAM needs, and how to share models with A1111.

  5. First image (text to image)

    Generate your first image with ComfyUI's default text-to-image workflow. Learn six nodes, their settings, batch generation, and fixed seeds.

Core techniques

  1. KSampler parameters

    Learn what each ComfyUI KSampler parameter does and how to choose seed, steps, CFG, sampler, scheduler, and denoise settings, plus when to use KSampler Advanced.

  2. Prompting techniques

    Write better AI image prompts with prompt structure, negative prompts, ComfyUI weight syntax, model-specific tips for SD 1.5, SDXL, and Flux, and useful English keywords.

  3. Image to image (img2img)

    Use ComfyUI image-to-image to create new images from a source image. Learn Load Image, VAE Encode, denoise, resizing, and practical style transfer workflows.

  4. Inpainting and outpainting

    Learn to edit image regions in ComfyUI with masks, compare inpainting workflows, use dedicated inpainting models, and extend image borders with outpainting.

  5. LoRA

    Learn what LoRA is and how to use it in ComfyUI: download and install LoRAs, connect Load LoRA, tune strength_model and strength_clip, use trigger words, stack LoRAs, and troubleshoot common issues.

  6. ControlNet

    Precisely control AI image composition with ControlNet: Canny edges, OpenPose poses, Depth maps, preprocessors, Apply ControlNet settings, strength, start and end times, and multiple ControlNets.

  7. Image upscaling

    Upscale AI images to high resolution in ComfyUI with upscaling models such as ESRGAN, latent upscaling, high-resolution fixes, and tiled upscaling.

Advanced Topics

  1. Custom Nodes and Manager

    Learn to install and manage ComfyUI custom nodes with ComfyUI-Manager, find missing workflow nodes, update or disable nodes, install manually, and choose popular nodes safely.

  2. Flux and other new-generation models

    Use Flux, Qwen-Image, and other new-generation image models in ComfyUI. Learn how they differ from SDXL, where to place separate model files, how to start with official templates, and how to choose quantized versions for your GPU memory.

  3. AI video generation

    Generate AI videos with ComfyUI. Learn text-to-video and image-to-video, open source models, official templates, frame and resolution settings, MP4 output, and hardware requirements.

  4. Automate with the API

    Use the ComfyUI API to generate images with code: export API workflows, submit jobs with POST /prompt, edit prompts and seeds, check history and download images, track progress with WebSockets, and use a complete Python example.

  5. FAQ and troubleshooting

    Troubleshoot common ComfyUI errors, including CUDA out of memory, missing models or nodes, black images, slow generation, and startup failures.

Updated October 12, 2026