A recent Youtube Web update has added a canvas whenever the seek bar is visible, an HTML5 canvas pops up. This was not asked for and not needed. If you disable canvases for privacy, this will cause a horrific red bad to cover half the screen until you hide the seekbar. Canvases can be used for fingerprinting, which I’m sure Google is doing here.

  • Kairos@lemmy.today
    link
    fedilink
    English
    arrow-up
    29
    arrow-down
    1
    ·
    8 days ago

    Google uselessly uses the canvas for its reverse image search. And I do mean uselessly - The image you upload is put onto the canvas, then immediately relayed to the server and never used again.

    • Celestus@lemm.ee
      link
      fedilink
      English
      arrow-up
      30
      ·
      8 days ago

      This is probably a clever way of doing native JPEG image conversion on the front end, instead of pulling in (or reimplementing) a universal image conversion library

        • purplemonkeymad@programming.dev
          link
          fedilink
          English
          arrow-up
          2
          ·
          6 days ago

          It’s probably more of a scale thing, going a conversation server side need CPU time, if it can be done prior to upload then server time is reduced. I think a lot of websites do client side processing so they can do more requests per server instance.

        • kevincox@lemmy.ml
          link
          fedilink
          English
          arrow-up
          1
          ·
          6 days ago

          It would be wasteful to upload the full size image only to throw most of it away. JPEG compression is very cheap, especially at low resolutions (I assume that image search uses a pretty low-resolution source image). Doing it this way is actually what I would do for best user experience. (Not saying that they aren’t doing other malicious things, but doing the resizing on the client is actually a good idea)

        • Celestus@lemm.ee
          link
          fedilink
          English
          arrow-up
          23
          ·
          8 days ago

          Yeah, I bet it would be trivial for one of their engineers to whip up a universally compatible, hardware accelerated image file converter in JS, using no external dependencies, and less than 50 lines of code. Hint: it uses Canvas

              • Kairos@lemmy.today
                link
                fedilink
                English
                arrow-up
                1
                arrow-down
                7
                ·
                7 days ago

                LibreWolf

                TinEye seems to have no problem with this. It seems weird to argue that something that doesn’t work on every browser should be used because the alternative doesn’t work on every browser.

                Google could easily do both. Using JS if canvas fails.

                • lime!@feddit.nu
                  link
                  fedilink
                  English
                  arrow-up
                  13
                  ·
                  edit-2
                  7 days ago

                  librewolf has canvas turned off, because it’s fingerprintable. it’s still in the firefox codebase. all major browsers support canvas and have for more than 10 fifteen years.

                  also, canvas is literally a JS API what are you talking about