A few years ago someone posted this on Reddit claiming to have made this character in blender. I thought that was really fucking dumb because it would take less than an hour and it was clearly ai. They argued with me when I said as much, so I did it myself in about 30 minutes to prove a point.

  • ObituarykidneyOP
    link
    fedilink
    English
    arrow-up
    78
    ·
    1 day ago

    Viewport preview to prove I have the 3D models made, that the original “artist” was unable to provide.

    • ObituarykidneyOP
      link
      fedilink
      English
      arrow-up
      81
      ·
      1 day ago

      Bonus render with the visor up that the op couldn’t do either.

    • Rhaedas@fedia.io
      link
      fedilink
      arrow-up
      6
      ·
      1 day ago

      A few years ago that was a given. I believe that you did that yourself. But AI frontier models, or some of them, can generate blend files now. They are probably not as good as a person who knows Blender well, and they still can’t do the more complex texturing and such. But they might be able to do the base model.

      Now, it wouldn’t take the 30 mins they took to use AI to do the first one, lots of edits and redos to get things right. To do a lot of models, it’s best to just learn Blender. Hell, you might learn Blender by having to help the model correct mistakes. But they can do some basics now.

      • ObituarykidneyOP
        link
        fedilink
        English
        arrow-up
        18
        ·
        1 day ago

        You can get chat gpt to use blender now too. Watched a video yesterday where you can just straight up tell chat gpt what you want and it feeds python scripts into blender that do the modelling. So you can’t even use wireframes or alternate angles to prove its in blender now.

        But it took the YouTuber 90 minutes and like $20 of tokens to do one scene. And typical to ai it looked great at first glance but on closer inspection things were weird and required fixing. Like cutlery on the table was rotated wrong and floating, chairs were clipped into the wall, window panes were beveled an insane amount but it was in the frame so it’s just wasted geometry.

        And for a person who’s never used blender the UI and navigation is confusing, so it’s still not as accessible as saying their ai slop is a drawing lol. Having the ai render out a fake wireframe or blender viewport is still hard to get right because it requires an even grid or a lot of text, so an experienced blender user could still tell on a closer inspection.