English
ComfyUI Desktop is a standalone application developed by the official ComfyUI team, compatible with Windows and macOS systems (partially supported Linux). It ports the original Web-based ComfyUI experience to the desktop environment, aiming to provide users with a more stable and intuitive AI content generation platform. Unlike traditional Stable Diffusion interfaces (such as Automatic1111, A1111 for short), ComfyUI adopts a unique flowchart-style node system, where users build custom workflows by connecting to "nodes", and achieve high control over the generation process without programming knowledge.
ComfyUI Desktop was officially released on October 21, 2024 and was open sourced on November 26, 2024. It simplifies the installation process and optimizes the user experience, making it ideal for AI art creators and developers.
The core of ComfyUI is its flexible node-based workflow, allowing users to design and perform complex AI content generation tasks. Here is an overview of the main features:
Basic functions: from text generation image (Text-to-Image), to image editing (internal complementary image, external expansion image), super resolution enlargement, etc.
Advanced features: Supports ControlNet (precise control image generation), LoRA (low rank adaptation), IPAdapter (image style migration), and even video generation through AnimateDiff or Stable Video Diffusion.
Model support: compatible with a variety of diffusion models, including Stable Diffusion (SD1.x, SD2.x, SDXL), Flux.1, etc.
Customization: Each node represents a functional module (such as loading a model, encoding prompts, and generating images), and users can freely combine it to meet their needs.
Whether it is simple tasks or complex experiments, ComfyUI can provide an efficient and transparent operation experience.
The official download channels for ComfyUI Desktop are as follows:
Visit address: www.comfy.org
step:
Windows User: Download the installation package for NVIDIA GPU .
macOS users: Download the version for Apple Silicon .
Enter the homepage of the official website or download page.
Select version according to the operating system:
After the download is complete, get a compressed file (such as .zip or .dmg format).
Visit address: github.com/Comfy-Org/desktop
step:
Go to the "Releases" page of the warehouse.
Find the latest version and download the installation package that matches your system.
Advanced users can clone the repository and build it manually (reliances such as Node.js and Python are required to install dependencies).
Note: As of March 5, 2025, it is recommended to prioritize the use of prebuilt installation packages for stability.
Here are the detailed steps to install ComfyUI desktop:
Start the installation:
Windows: Unzip the .zip file and double-click ComfyUI Desktop.exe .
macOS: Open the .dmg file and drag the icon to the "Applications" folder.
Select the path:
The installation path must be in all English, avoid Chinese characters, and the disk space is no less than 15GB.
File Migration (optional):
If you have installed the old version of ComfyUI, you can import existing models and settings; new users can skip them.
Complete the installation:
Click "Install" and the system will automatically download the dependencies. After completion, generate a desktop shortcut icon and double-click to start.
Manual import: Copy the model files ( .ckpt , .safetensors , etc.) to the corresponding subfolder in the ComfyUI/models directory.
Automatic import: Supports automatic download through model links in the workflow.
Built-in ComfyUI Manager, users can search and manually install custom nodes in the plug-in interface.
Click the "Bottom Panel" icon in the upper right corner of the interface to view the model download progress, generation status and error prompts.
Update: Automatically detect new versions at startup, and update options can be adjusted in Settings.
Uninstall: Delete the home folder and clean the remaining files in the system.
To ensure normal operation, the equipment must meet the following minimum requirements:
Windows: Windows 10 or later, NVIDIA GPU (4GB or more video memory is recommended).
macOS: macOS 11 (Big Sur) or later, Apple M series chips.
Storage space: At least 15GB of available space (including model files).
Network: The first time you run it, you need to download and dependencies online.
Compared with Automatic1111, ComfyUI has the following advantages:
Flexibility: Node-based design provides higher customization capabilities.
Efficiency: Generate speed is 3-5 times faster, and only changes in the workflow are re-execute.
Low video memory support: Minimum 1GB of video memory can run (higher video memory is recommended).
Although its learning curve is slightly steep, ComfyUI is an irreplaceable option for users who pursue deep control and efficiency.
Nodal workflow: highly customizable, such as text-to-image flows including loading models, CLIP encoding, potential image generation, and VAE decoding.
Efficiency: Optimize performance and support complex tasks.
Model compatibility: supports multiple formats ( .ckpt , .safetensors , .diffusers ) and independent VAE, CLIP, etc.
Scalability: The community provides rich custom nodes, supporting video generation, 3D modeling, etc.
Shareability: Workflows can be saved as JSON or embedded with PNG metadata for easy sharing.
When you encounter problems, you can seek help in the following ways:
Official forum: forum.comfy.org
GitHub Issues: github.com/Comfy-Org/desktop/issues
Community Communication: Join the ComfyUI community and interact with other users.
ComfyUI desktop is a powerful AI creation tool that combines node-based workflows and the convenience of desktop applications to provide users with an efficient and flexible content generation experience. Visit www.comfy.org or github.com/Comfy-Org/desktop now to start your AI art journey!
