Only 7 pics in training data? I think there's more pics of her in her book than that (and she makes a few more token appearances in the later Spider-Verse stuff). I have enough experience with comic book character LoRAs to tell you not to be afraid comic artifacts (speech bubble, panel divider etc.)) or partial shots, as long as you tag them right ("comic", speech bubble, blank speech bubble, portrait etc.) and blank the speech bubbles.