Feature Request - Text Mask, Masked Crop nodes

For refining specific portions of a generated image, it would be nice to be able to ask clipseg what it thinks the mask should be for that, with some optional padding/tolerance- then I can run img2img and paste the modified segment back into main image.

There's an A11111 extension that already does this, txt2mask- the initial release commit is under MIT, not sure about the current one as the author merged it into Unprompted.