The AI image generation software ComfyUI has been released in version 1.0 and now offers a desktop version for Windows, MacOS and Linux. Background: ComfyUI allows you to implement complex image generation pipelines directly on your own computer. Unlike cloud-based solutions, the software provides full control over the process and your own data.
The brand new desktop version is currently in closed beta and offers automatic updates, an improved user interface with tabs and custom keyboard shortcuts, and an integrated log viewer. Another new feature is automatic model download, which allows you to embed models directly into your workflow and download them on demand. A new model browser simplifies model management. The new interface is already included in the current version of ComfyUI and can be activated in the settings.
ComfyUI itself has a modular structure and uses “nodes” to connect different image processing steps. You can think of them as building blocks that are connected to achieve a complex result. Each node has a specific function, such as loading an AI model, applying a filter, or adjusting parameters. Connecting the nodes creates a directed graph that defines the image generation process. Data flows from node to node and is transformed step by step. This modular structure provides a high degree of flexibility and control over the entire process. Users can use predefined nodes or create their own to design custom workflows.
Version 1.0 introduces the Comfy Node Registry (CNR), a central repository for user-defined nodes. The desktop application will be the first to support installation from the CNR.
The standard web version of ComfyUI will continue to be developed in parallel. The desktop app is ultimately a wrapper around this web application and makes it easier to use ComfyUI, especially for non-technical users. The source code of the desktop app will be released in the coming weeks as part of an open beta phase.