• @[email protected]
    link
    fedilink
    English
    367 days ago

    Excuse me, I forgot the text. Wouldn’t be fair without all that entropy. To make up for it, I turned on chroma subsampling. This one is 39 KiB.

    • 🇰 🌀 🇱 🇦 🇳 🇦 🇰 🇮
      link
      fedilink
      English
      21
      edit-2
      7 days ago

      To be more accurate, the dithering was more specific to animated gifs and not static pictures. Unless you go back really far… I remember finding porn on 5 inch floppies my dad had in his closet and even the non-animated images were black & white and dithered to hell. Though I am pretty sure he didn’t obtain those off the internet.

      • LostXOR
        link
        fedilink
        376 days ago

        Depends on what you consider readable, but I got it down to 2842 bytes by downscaling it and using an absolutely atrocious quality setting:

        Image

        That’s just small enough to fit inside a QR code, so I did that! (You probably can’t scan it with a typical QR reader; you need something like zbar that supports reading binary data).

        • @[email protected]OP
          link
          fedilink
          126 days ago

          nice :D

          in a different timeline where the GIF89 specification would not have been mostly ignored, it would have been possible to go even smaller.

          “The GIF89 specification allows you to specify text captions to be overlayed on the following image. This feature never took off; browsers and image-processing applications such as Photoshop ignore it”

          Meaning if your gif viewing client supports full GIF89 then you could just display the text over a 1x1 pixel image, shrinking the file size down to something in the range of < 100bytes.

          • LostXOR
            link
            fedilink
            16 days ago

            At that point you might as well just send it as text.

          • GreatAlbatross
            link
            fedilink
            English
            26 days ago

            In that vein, I’d be interested to see if someone could Inkscape it into an SVG with embedded text.

            Though I have no idea how many vectors would be needed to make the image similar enough.

            • @[email protected]OP
              link
              fedilink
              16 days ago

              The SVG send me down a rabid hole, apparently SVG embedded text is actually selectable and all, this is basically what gif tryed to do and failed lol

              but sadly lemmy dosent seem to let upload .svg :/

          • LostXOR
            link
            fedilink
            16 days ago

            Unfortunately most QR readers don’t recognize an image regardless of the data format.

    • @[email protected]
      link
      fedilink
      English
      6
      edit-2
      6 days ago

      I turned on chroma subsampling

      A simplified polygonal scene, originally from Futurama scene that's styled like anime. Fry is saying "You and I are enemies now." while pointing at a jpegified Professor Farnsworth. Meta note: The scene was made in the Godot game engine.

      43KiB, though obviously a chunk of that is from the colors of the jpg Professor.

      EDIT: I just noticed webp messes with colors around the edges too (which would matter more if I cut the resolution in half or so). The original .png (what I actually uploaded) doesn’t at 51.6KiB.

      Also obviously this would be less data in non-pixel format (well, it’d likely be eaten up by overhead for packaging, though after that it would allow tons of art for negligible data cost). Would be smaller as an svg for example (or an swf).

    • @[email protected]
      link
      fedilink
      46 days ago

      Dithering is much older than jpeg.

      They used dithering in printed media, so it was transfered to digital media as a way to save on color depth, before jpeg got invented.