All the power of ComfyUI.
None of the node graph.
Make images. Not node graphs.
MooshieUI: Generate 
Everything you generate, none of the graph-wrangling
MooshieUI keeps ComfyUI's power and hides its complexity behind controls that make sense.
Guided, not wired
Prompts, dimensions, samplers, and models are laid out as plain controls. Switch settings without ever touching a node or a connection.
Three generation modes
Text to image, image to image, and inpainting. Switchable with a single click, sharing the same prompt and model setup.
LoRAs & Artist Styles
Stack LoRAs with independent model and CLIP strength, and apply bundled artist styles that inject tags without cluttering your prompt.
Built-in refiner
Tiled-diffusion upscaling with OmniSR. Every tile is denoised simultaneously each step for seamless, high-quality results.
Compare side by side
Generate batches and line results up in a compare grid to find the seed, sampler, or prompt tweak that actually worked.
Works on your phone
The browser build is mobile-aware, so you can kick off and check on generations from any device on your network.
Same interface, desktop or browser
MooshieUI ships as a native desktop app and a self-hostable web server. Pick whichever fits your setup. The UI is identical.
Install & launch
A native app built with Tauri and a Rust core. Fast to start, light on memory, no Electron bloat. Windows and Linux ship as ready-to-run builds; macOS builds from source.
- Native window, system tray, and file dialogs
- Tiny footprint thanks to the Rust backend
- Bundles ComfyUI with one-click setup
Self-host & share
Run the built-in web server and reach MooshieUI from any device on your LAN. Docker-friendly and mobile-aware. Perfect for a GPU box in the closet and a laptop on the couch.
- One server, many devices over your network
- Docker-ready for a clean, repeatable deploy
- Responsive UI that adapts down to phone screens
Three steps to a finished image
- 1
Describe it
Type a positive and negative prompt, or apply quality prompts and artist styles. Pick an aspect ratio and resolution from presets.
- 2
Dial it in
Choose a checkpoint, stack LoRAs, and set your sampler. Or hit “Apply” to load the model's recommended steps, CFG, and scheduler.
- 3
Generate & refine
Watch live progress, then send the result through the tiled refiner for a clean, high-resolution upscale.
Get started in minutes
Free and open source. Download the installer and run it. The setup wizard installs ComfyUI, Python, and PyTorch automatically on first launch.
Windows
One-click installer. The setup wizard installs ComfyUI, Python 3.11, and PyTorch on first launch. No manual configuration needed.
Download .exeLinux
AppImage runs anywhere without installation. The .deb integrates with apt on Ubuntu/Debian-based systems.
macOS & Docker
macOS builds from source in a few steps. For a self-hosted server on any OS, one docker compose up is all it takes.