SdPaint is a desktop painting application that puts Stable Diffusion inside a familiar canvas-based paint program. Instead of typing prompts into a web form, artists paint, sketch and mask directly on a canvas and let the diffusion model generate, fill and transform regions of the artwork in place. It is an open-source project on GitHub with around 1,600 stars, written for Windows desktop use.
The application pairs a traditional paint interface – brushes, layers of strokes and selections – with a Stable Diffusion backend, so generation becomes part of the drawing process rather than a separate step. Users can sketch a rough composition and have the model render it, paint over generated areas to guide revisions, and iterate visually in both directions. Because it runs locally against a Stable Diffusion installation, there are no per-image fees and work stays on the artist’s own machine.
SdPaint is free and open source under the MIT license. You supply your own Stable Diffusion setup and hardware.
Digital artists and illustrators who want AI generation woven directly into a hands-on painting workflow on their own machine.
SdPaint belongs to the first generation of tools that treated diffusion models as a medium to paint with rather than a slot machine for finished images. Because generation happens on a canvas the artist controls, the workflow encourages happy accidents in both directions: rough human strokes guide the model, and unexpected model output can be painted over, masked and regenerated until the piece comes together, much like working in a traditional layered painting program.