based on your example prompts, does that mean you trained this using tagging instead of captions?

I have been trying to do this with prompts for a month to no avail!
Thank you - this is exactly right

Great lora. With selfie tag on your favorite model (does not need a lora for selfie, just type "selfie"), this one works really well because it captures the dudes directly behind. Just be creative, like the dudes are looking at you, indicating your phone. Make sure put "on all four" and "low angle" and "she is holding the phone" and a "true selfie." You get 99% of the time dudes on a train, giving her what she wants hehe. ;)