Hi!

This is actually my first post on Lemmy so sorry if anything is messed up, but I was just wondering if anyone’s explored (or knows of anyone who has) the idea of training a stable diffusion dreambooth/LoRa on their face then asking it to feminize it.

I’m not currently in the position to easily or openly explore too much, and thought this might help me personally (and if no one else has I might try it myself then release the steps I took!)

  • apprehentice@lemmy.enchanted.social
    link
    fedilink
    arrow-up
    31
    ·
    1 year ago

    I’ve heard that this can be harmful because it would give you unrealistic expectations. An AI image model would grossly exaggerate feminine features in this case, more so than would become apparent on gender affirming treatment. Male to female camera filters are the same way. I would steer clear of this train of thought, personally.

    • Sethayy@sh.itjust.worksOP
      link
      fedilink
      arrow-up
      4
      ·
      1 year ago

      That’s actually a really good point I didn’t think of, I wonder if you could train it on before/after images to get something more modest and bone-shape-respecting