MS Paint meets Stable Diffusion, with 47 keyboard shortcuts

What it does SdPaint is a Python painting app that pipes your canvas directly to the Automatic1111 Stable Diffusion WebUI. Every brushstroke gets shipped via API to ControlNet (scribble/lineart models), and the generated image bounces back to your canvas. It also offers a web interface and an experimental img2img mode that watches a file for changes.

The interesting bit The project treats Stable Diffusion as a real-time rendering engine rather than a batch tool. The “quick mode” drops to 6-step LCM LoRA generation for near-instant feedback, and the keyboard-heavy workflow (cycle samplers, denoising strengths, HR upscalers mid-sketch) suggests someone actually uses this for iterative drawing, not just prompt engineering.

Key highlights

• Live ControlNet integration: scribble/lineart detection with switchable models and detectors
• Quick mode with LCM LoRA support for low-step generation (6 steps, DPM++ SDE Karras by default)
• Extensive hotkey workflow: presets, batch rendering, seed control, autosave, all keyboard-driven
• Web interface available via StartWeb.bat
• Custom JSON configs for prompts, ControlNet weights, and rendering presets

Caveats

• Linux installation is listed as “To Do” — Windows and macOS only have documented setup
• Requires a running Automatic1111 WebUI with ControlNet extension, API mode enabled, and specific model downloads
• “Pixel perfect” preprocessor disabled by default because it reportedly hurts quality above 512px

Verdict Worth a look if you want to sketch loosely and let AI handle the rendering pipeline in real time. Skip it if you need a polished standalone app or don’t already have Automatic1111 running locally.