• loaExMachina@sh.itjust.works
      link
      fedilink
      arrow-up
      42
      ·
      14 days ago

      OpenAI released a generative model that emulates the famous Studio Ghibli art style; or more precisely Hayao Miyazaki’s art style. The latter had once said “I strongly feel that this is an insult to life itself” when presented with a proof of concept of AI generated 3D characters, an exchange you can watch here. Now, this specific sentence refers more to the idea of using unnatural movements similar to a disability for creepy shock factor; but he also seems avert to the concept of AI art.

      • kautau@lemmy.world
        link
        fedilink
        arrow-up
        22
        ·
        14 days ago

        Also to add context for those unaware, this is meant to be JD Vance, the US Vice President

      • Pennomi@lemmy.world
        link
        fedilink
        English
        arrow-up
        10
        ·
        14 days ago

        I’m not shocked that the oldest generation of animators are anti-AI art. The quality isn’t professional yet (even with the newest models) but it’s just good enough now that it’s easier to slap together something with AI than to pay a real artist. Quality vs cost/time tradeoffs are hard for real professionals to accept.

        • Timecircleline@sh.itjust.works
          link
          fedilink
          arrow-up
          7
          ·
          13 days ago

          Hayao Miyazaki’s art style specifically has so much life injected into it, that you would have to be incredibly specific to even approach his style.

          There’s an instance that stands out in Spirited Away, when Chihiro is putting on her shoes, she taps the toes of one foot against the ground to set it on correctly. A miniscule detail, but one that makes her feel more real. You’d need such an elaborate prompt to make that happen via AI, but thats the kind of detail that he includes in his movies. To him, the thought of having “someone” illustrate that who has never put their shoe on incorrectly and had to fix it is beyond inauthentic. That’s my take. Less about the worry of technology taking over, and more about the technology not understanding the language it’s being taught, because it can’t.

          • Pennomi@lemmy.world
            link
            fedilink
            English
            arrow-up
            2
            ·
            13 days ago

            Animation is a totally different beast than still images, of course. Almost all AI video is complete shit right now.

        • ZILtoid1991@lemmy.world
          link
          fedilink
          arrow-up
          1
          ·
          10 days ago

          There were some AI models that could interpolate animation frames, specifically made for animators.

          However, all trace of it disappeared and development resources were instead concentrated on generating animations from a few text prompts, primarily made for non-artists like corporate executives and investors.

          • Pennomi@lemmy.world
            link
            fedilink
            English
            arrow-up
            1
            ·
            10 days ago

            Modern AI video models do frame interpolation too, at a quality far superior to old AIs. It’s not advertised as heavily but it’s definitely a capability that is directly useful for traditional animation studios.