Announcement

Collapse
No announcement yet.

Could JPEG2000 Finally Take Off In 2020? It's A Possibility With High Throughput HTJ2K

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #31

    Comment


    • #32
      Originally posted by sandy8925 View Post
      Faster decoding and encoding is highly beneficial for mobile device.
      Depends how you look at it. For most mobile devices with a sensor of a sensible pixel density, the transcoding is plenty fast enough. And I say "sensible" because 25MP+ on an itty bitty sensor isn't sensible. All the advantages you get by having more pixels is lost due to extra noise (more pixels yields a darker image, and a darker image needs a higher ISO/gain, which adds noise). I imagine the rolling shutter effect is exacerbated by a higher pixel density too.

      Comment


      • #33
        I don't see JPEG2000 gaining popularity. We definitely need to do away with the 8 bit limit of JPEG, add larger color space, greater dynamic range etc but JPEG2000 ain't it. Canon has started putting HEIF in it's cameras due to the need to be able to move and store large numbers of images on a performance restricted platform like a camera. But the patent mess I think will mean most people will just be converting those to some thing else once they get them off the camera.

        Comment


        • #34
          Hmm... this is just an example of a technical solution to a problem that no one has.

          PNG and JPEG are “good enough”. If you’re delivering video, your bandwidth and storage is 100x what you need.

          There are plenty of other technologies that are “good enough” and are successfully fighting “technically better versions”: MP3, H264, PDF, Javascript, C, 16bit audio, 24bit RGB, USB, Windows 7.

          Comment


          • #35
            Originally posted by Marc Driftmeyer View Post

            No, it's a sign that Apple is willing to contribute some IP to interoperate with HEIV/HEIC/HEIF. Apple won't abandon HEIC as it has proven quite invaluable already in the macOS/iOS/tvOS/WatchOS ecosystem.
            Except that no streaming service is using it. They're all sticking with the decades old H.264 due to the hopeless patent quagmire of H.265.

            Only a few adopted VP9 (Google, Netflix, and now Hotstar).

            Comment


            • #36
              Originally posted by OneTimeShot View Post
              ​​​​
              There are plenty of other technologies that are “good enough” and are successfully fighting “technically better versions”: MP3, H264, PDF, Javascript, C, 16bit audio, 24bit RGB, USB, Windows 7.
              Don't forget HDMI. DisplayPort is just sooo much better.

              Comment


              • #37
                Originally posted by DebianLinuxero View Post
                Does anyone here remember MP3 PRO?

                This is the same thing.
                This comparison makes no sense. MP3Pro was a gimmick tech which was not accepted since it was patented, required fees and was basically a slightly improved MPEG2 Layer 3 audio codec. JPEG2000 has almost nothing in common with JPEG except both are used to compress images.

                Comment


                • #38
                  Originally posted by sandy8925 View Post

                  Don't forget HDMI. DisplayPort is just sooo much better.
                  To be fair, DisplayPort gets those advantages by being a significantly more complex thing to implement, so I can see why there'd be reluctance to spend that effort.

                  Comment


                  • #39
                    Regarding "We definitely need to do away with the 8 bit limit of JPEG" - for people interested in high-quality still or moving image production, yes, sure.
                    But 99.9% of all uses of image formats are done by users who could not care less about color spaces and high dynamic range. Realistically, JPEG is "good enough" for almost everyone, just as MP3 is "good enough" for almost everyone. JPEG is here to stay indefinitely, with other formats competing for a 0.1% use case residual.

                    Comment


                    • #40
                      By the way, has anyone actually looked at the whitepaper? IMHO it's actually less impressive than the initial marketing suggests.

                      Performance: it *is* much faster, no doubt. But only in relation to the original JPEG2000! HTJ2K is about the same throughput as original JPEG. They compare their multithreaded HTJ2K implementation against a singlethreaded JPEG implementation, which isn't apples-to-apples.

                      Compression ratio: the new block coding method introduced with HTJ2K actually has significantly worse coding efficiency. File sizes for the same quality increase by 6-7%. So the compression ratio advantages of JPEG2000 (if any) mostly evaporate.

                      My conclusion is that HTJ2K might turn out to be great for the few niches where JPEG2000 is already established, but there's not much chance it will see widespread adoption.

                      Comment

                      Working...
                      X