Afaik no one knows where the line between “human-made” and “machine-made” is, so AI-generated imagery is considered public domain (at least for now).

    • Mossy Feathers (She/They)@pawb.socialOP
      link
      fedilink
      arrow-up
      0
      ·
      2 months ago

      Honestly, that’s a reasonable take. I mean, there’s a world of a difference between using an AI to automatically unwrap and generate UVs or generate PBR maps from flat images vs having an AI auto-generate textures and models from scratch. What’ll be interesting is to see where the line ends up being drawn.

    • rtxn@lemmy.world
      link
      fedilink
      English
      arrow-up
      0
      ·
      2 months ago

      it’s okay, just tell the AI to reproduce that image one-to-one. The result definitely doesn’t have enough human authorship to qualify for protection.

      • Even_Adder@lemmy.dbzer0.com
        link
        fedilink
        English
        arrow-up
        0
        ·
        2 months ago

        You’re good to go as long as it’s original enough. If it isn’t then that’s just copyright infringement. You might as well have right-clicked “save as”.

      • Mossy Feathers (She/They)@pawb.socialOP
        link
        fedilink
        arrow-up
        0
        ·
        edit-2
        2 months ago

        The problem with that is that copyright is viral. If you take a photo of a work under copyright, then your photo is also under that copyright. You own the copyright on the photo, but the original copyright holder owns the rights for the reproduction of the copyrighted work; meaning you both own copyright on the photo.

        It usually isn’t enforced, but you can see an example of this when news stations get sent C&Ds or DMCAs because some chucklefuck with a loud stereo decided to harass them (at least, I’m pretty sure I’ve heard of news stations getting legal letters as a result of someone playing unlicensed music in the background).

        To make matters worse, because anything can be considered artistic with a good enough story, you can potentially claim copyright on reproductions of public domain works if you alter them in some way; even if that’s just by changing the spelling of some names.

        “Fun” fact: this is why some museums try to stop you from taking photos. The artifacts and paintings being presented may not be under copyright anymore, but the museum is claiming copyright on the presentation of the artwork or artifact, not the artwork or artifact itself. Afaik it’s never gone to court so no one knows if it’d actually hold up (I think they usually cite other reasons for kicking you out, which makes it harder to fight), but you’d have to go to court to find out and there’s no penalty for them falsely claiming copyright.

        All that said, however, it doesn’t take much for it to be considered derivative.

        Yet, still, just putting the image into a bot and tell it to make a copy wouldn’t be enough to free it from copyright; you just can’t claim copyright on the (unauthorized) reproduction. The bar is pretty low unless it can be proven that you intended to infringe on someone’s copyright (putting a image through a bot with the express purpose of getting a “copyright-free” copy would qualify), or even just awareness of someone’s work, even if you didn’t mean to copy it, can get you fucked.

        There’s been a fun case or two recently where musicians got sued because their melodies were the same as someone else’s and the company was able to successfully argue that the original melodies were high-profile enough that said musicians must have been aware they were copying a melody.

        Copyright is so fucked. It’s not a bad idea, but holy shit is its modern implementation fucked.

  • Swedneck@discuss.tchncs.de
    link
    fedilink
    arrow-up
    0
    ·
    2 months ago

    also make sure to post them online while pretending it’s made by a human, so future models are trained on it and slowly get more and more insane as more of their own output is fed back to themselves as training data.