AI image editing felt like testing my ability to explain what I see: “exactly X, preserve Y, the person on the Z..." and it would still edit the wrong stuff.

So I built a paint-and-prompt editor.

(1) You paint the region

(2) Type a short prompt, and done

Prompt before: "Remove the person in the far-left corner behind the person wearing a black jacket. Preserve everything else in the image, no identity change... etc"

Prompt now: "Remove him."

Website: https://swisy.app

If you try it, it will break. Please do let me know what sucked and why not worth trying. I'll fix it.