SD WebUI Segment Anything is an extension that brings Meta’s Segment Anything Model (SAM) into the Stable Diffusion WebUI, adding precise, promptable segmentation to the image-generation workflow. With it you can click points on an image to select exactly the region you mean, then inpaint, edit or composite only that region — a level of control that text prompts alone cannot give. The extension has around 3,500 stars on GitHub.
The extension runs SAM locally inside the WebUI process and exposes interactive point-and-box prompting directly on the canvas. Masks can be refined, combined and sent to img2img inpainting, ControlNet workflows or external editors, and the extension integrates with other popular WebUI tooling to automate steps such as background removal and object isolation. It supports the original SAM as well as faster variants, with model files downloaded on first use. For artists who iterate on generated images, being able to say “this object, exactly” — and get a clean mask in one click — removes one of the most tedious parts of the pipeline, manual masking.
The extension is completely free and open source; it runs locally inside your own Stable Diffusion WebUI installation, so the only cost is your GPU time.
Stable Diffusion artists and image editors who need accurate masks without leaving the WebUI — for targeted inpainting, product-image cleanup, background swaps and other region-precise edits. Note it is an extension that performs its own SAM inference, not a settings companion.