Good afternoon!
I'm a little late with this workflow)). Now that Flux.2 Klein has been released and z-image-edit is coming soon, SDXL will quickly become obsolete. So it's better to release it now while it can still compete.
What's this workflow for?
It generates clothing for a character based on your sketch and prompt, and then uses masks created by Sam3 to separate the clothing into a separate image. I wanted to create a tool for creating clothing assets for visual novels.
Attention!
Unfortunately, this doesn't work perfectly - sometimes a little skin can show through the clothing. So you'll probably need to refine the result in Photoshop or Krita.
Description
Fix 1.0.1:
Replaced nodes for SAM3 with those built into ComfyUI
FAQ
Comments (2)
I'm having trouble with this, not sure how to do this without a sketch
Just draw a mask in the mask editor and write a prompt, then increase the noise in "ImageAddNoise".







