• Fisk400@feddit.nu
    link
    fedilink
    English
    arrow-up
    67
    arrow-down
    3
    ·
    1 year ago

    The developers probably added a pretty detailed prompt to the backend that the AI uses whenever someone asks it to do this. If I was making an AI, I would not risk it spitting out something scary when given this prompt.

      • leonardo_arachoo@lemm.ee
        link
        fedilink
        English
        arrow-up
        28
        ·
        1 year ago

        While cool, I am surprised this is considered noteworthy enough for an article. There’s so much experimentation constantly going on with these models, what makes this special? It’s more like the inspiration for a creepypasta.

      • Just_Pizza_Crust@lemmy.world
        link
        fedilink
        English
        arrow-up
        25
        arrow-down
        1
        ·
        1 year ago

        I’m a pretty firm believer that Loab is a hoax “cryptid” due the unwillingness of the author to publish anything on the generation parameters. I think the author got one weird result from putting “Loab” in the negative prompt and then used img2img from there. Anyone who shows images of a “cryptid” they discovered but refuses to show proofs is untrustworthy in my opinion.

    • Dizzy Devil Ducky@lemm.ee
      link
      fedilink
      English
      arrow-up
      20
      arrow-down
      1
      ·
      1 year ago

      If I was in charge of one of these large AIs, I’d totally have it create the opposite of what OP got. I’d have it be some random eldritch horror creature any time anyone asks it to draw itself just for the fun of it.