DiffusionDesk wraps ComfyUI in native desktop frontend for inpainting
A new open-source desktop app bridges ComfyUI's node power with streamlined inpainting UI, targeting Apple Silicon users tired of Python script workflows.
A developer released DiffusionDesk this week, a desktop frontend that runs ComfyUI workflows under the hood while exposing a cleaner inpainting interface. The project targets users who want ComfyUI's flexibility without wrestling with its node graph for every edit. The initial release supports local image generation, model management, and Apple Silicon optimization.
The creator describes the tool as sitting between ComfyUI's raw capability and the ease-of-use bar set by dedicated inpainting apps. The GitHub repository includes a dashboard that organizes assets, tracks prompt history, and manages models through a native UI layer.
What stands out
- 01Backend stays ComfyUI. DiffusionDesk doesn't replace the node system — it wraps it. Users who already run ComfyUI locally can point the frontend at their existing install.
- 02Inpainting as the entry point. The interface prioritizes mask-and-fill workflows over txt2img, a design choice aimed at users who spend more time editing than generating from scratch.
- 03Apple Silicon first. Native M-series support ships in the initial release, a nod to the Mac-heavy local-gen community that often waits months for optimization.
- 04Asset organization built in. The app includes a media library and prompt tracker, features that typically require third-party tools or manual folder management in A1111 and ComfyUI.
- 05Open roadmap. The developer is soliciting feedback on missing features and pain points in current tooling, signaling active iteration rather than a one-time release.