Does he realize anyone can make them? Like, for example, an image of him crying on his way to court while protesters throw cheese at him.

  • gedaliyah@lemmy.world
    link
    fedilink
    arrow-up
    52
    arrow-down
    2
    ·
    3 months ago

    Wow, AI images are getting really good. Everyone has the right number of fingers, the signs contain words made up of real letters, and are appropriate to the context… I had to look closely, even knowing that it was AI.

    • hperrin@lemmy.worldOP
      link
      fedilink
      arrow-up
      25
      ·
      3 months ago

      I know, right! I used a model called Flux to make this one. There were some obvious mistakes that I fixed by inpainting (regenerating a specific part of the image), but even the first version was really good.

    • magic_smoke@links.hackliberty.org
      link
      fedilink
      arrow-up
      18
      ·
      3 months ago

      Yeah but there’s still that weird play-doh softness to the people. Like the early subsurface scattering demos where everything looked like meat jelly.

      • Swedneck@discuss.tchncs.de
        link
        fedilink
        arrow-up
        1
        ·
        3 months ago

        that and the composition is just too perfect, AI generated stuff tends to suffer from what i can only call main character syndrome but in reverse

    • jawa21@lemmy.sdf.org
      link
      fedilink
      arrow-up
      8
      ·
      edit-2
      3 months ago

      Look again. The hand holding the “cheese the don” sign has an extra thumb coming from the back of it somehow. None of the fingers are proportioned correctly (like the woman on the left’s forefinger is much too long). There is a lot wrong.

      Editing to add the lady with 3 fingers and a thumb.