Past experiments developing "faceless dom" and "faceless sub" lora yielded something interesting- even when the tools failed at their intended purpose, they still managed to enhance the facial expressions and give the picture a stronger femdom "feel" as a result.

So why not take the training datasets for both, and use it as the starting point for a kind of instant "femdom vibes enhancer" people can add to any image?

Recommended starting triggers: "fdom_faces, blush"

Optional: "faceless male" and/or "faceless female". The former is definitely working better atm, chalk it up to available training data.