Announcement

Collapse
No announcement yet.

google is killing MPEG LA by droping h264:

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

  • #51
    Look, it's not about destroying h264 as a format, or a dick size comparison between h264 and vp8.

    What Q meant is that the steps from google are aimed at reducing the market share of h264, and stopping us/the net from getting even more addicted to h264. If that succeeds, it will significantly reduce the leeway for MPEG LA to screw us. And since the latter is it's raison d'ętre, 'killing it' that way.

    Am I good at interpreting Q or what

    Comment


    • #52
      Originally posted by evolution
      1- Content protection: Service and Content providers want something that guarantees their content is received and your receive it "protected". If Google can convince those companies their codec "protects" the contents the provide, there will be a huge bump of adoption of Google's VP8 codec.
      Video codecs don't protect anything, they compress digital video. H264 is as (in)secure as WebM.

      Originally posted by evolution
      2- Performance: ATM, VP8 is no better than H264 baseline in visual quality, and, in terms of performance is much slower than H264 (both encoding/decoding processes).
      I don't think you tried any recent ffmpeg build including ffvp8. Exact comparisons are hard for it's difficult to find two video samples that could be considered equivalent, but in general terms there are no big differences. The same x264 developer you no doubt took your comments about visual quality from (at least indirectly), could also tell you that "While there [is] no particular reason that [WebM] should be much faster than a good H.264 decoder, it shouldn’t have been that much slower either!". Google's official implementation also got quicker lately (it actually is an ongoing process), and the next (current?) release of libvp8 will focus on encoding speed.

      As for visual quality, it's not like there's a limit on how good a video can look like, you just throw more bits at it. So yes, WebM is no better than x264 at the same bit rate (although it probably is superior to the worse H264 implementations), but that's not the point. Obviously Google considers the codec to be good enough such as to not care having to deal with the little overhead its use represents over the best H264 implementation in the universe. I'd like to know whether all those people repeating what DS said in his infamous analysis actually did any tests of their own. I know I did, and the codec rocks. Sure it encodes slow, sure it doesn't surpass x264, but it really is good. Actually, if you don't use presets it's quite easy to screw up your x264 encoding and end up with a less-than-optimal size; why is everybody so concerned about this when most of the time the video content found in the web are nothing more than horrible encodings?

      Originally posted by evolution
      3- Hardware support: Now we'll have to wait some time until hardware manufacturers start to support VP8 along H264. And until then, we might have to wait some time...
      Right, and they are already starting.

      Originally posted by evolution
      4- For those who're insulting me saying I support patents: I also disagree with _CORPORATE_ patents, I think they stop technology progress in some ways... So, you're having a wrong opinion from me.
      See, I myself don't have a clear opinion on patents. I see good arguments coming from both sides, and I'm probably inclined to support patents for tangible products (like, not software). But this case is an easy pick, you've got two comparable products, one of them encumbered by patents to the bone, and the other patent-free unless proven otherwise. There's no need for deep philosophical arguments here.

      Originally posted by evolution
      5- Finally, for the "trolls": What I've been trying to tell since the beginning from this thread is I don't support the way Google is dropping H264 support because, nowadays, H264 videos are one of the most widespread used codecs for Internet / HD videos, and if they're doing this in the near future isn't a very good decision, because I think programs should be used based on its funcionality, and for me, ATM, Chrome is the best browser I can use on Linux.
      As a consequence, I think they should drop this support but in a later time...
      Furthermore, I don't agree with this decision because I think this will allow third-party _PROPRIETARY_ plugins (such as flash) to live longer, whereas with HTML5, most plugins you use can be replaced in the future by simpler "html tags".
      I can't understand you, sorry. I think you're arguing something around the lines of "everybody uses H264, so everybody must use it", or something. Be it as it may, I don't think this battle is lost, and Google people, with all the data they have available regarding browser market share and net usage, apparently agree.

      Comment


      • #53
        Originally posted by deanjo View Post
        Licenses should not apply to algorithms, mathematics and formulas as well. Could you imagine if Isaac Newton had a license on his contributions?
        That's called a red herring.

        You are perfectly aware that software patents are flat illegal in large parts of the world, because they patent mathematics, which is discovered, and not invented.

        Comment


        • #54
          Originally posted by pingufunkybeat View Post
          That's called a red herring.

          You are perfectly aware that software patents are flat illegal in large parts of the world, because they patent mathematics, which is discovered, and not invented.
          Umm software patents are not "illegal" anywhere. They are either recognized for validity or not.

          Comment


          • #55
            I might also add that software licensing is not recognized world wide by law either.

            Comment


            • #56
              Originally posted by deanjo View Post
              Umm software patents are not "illegal" anywhere. They are either recognized for validity or not.
              Fair enough, poor wording. Any such patent is not legally enforceable in much of the world.

              When it comes to licenses, your argumentation is confusing.

              Copyright is recognised worldwide. Copyright says that only the author of a piece of work has the right to distribute copies of said work.

              Software licenses are only needed because copyright exists. Remove copyright, and you don't need licenses. You just copy and use.

              If you want to abolish copyright, it's a lofty goal, but very much unrelated to any discussion in this thread.

              Comment


              • #57
                There shouldn't be anything confusing. Since algorithms, mathematics and formulas are not invented but discovered no one should have any say where or how they are used. A license is a form of "ownership" that allows the "owners" to dictate on the terms of use.

                Comment


                • #58
                  Algorithms are not hidden under a rock that when lifted the algorithm gets discovered. Algorithms are invented.

                  No idea about the other two.

                  Comment


                  • #59
                    Originally posted by deanjo View Post
                    There shouldn't be anything confusing. Since algorithms, mathematics and formulas are not invented but discovered no one should have any say where or how they are used. A license is a form of "ownership" that allows the "owners" to dictate on the terms of use.
                    And there is nothing confusing. I can freely implement a fast Fourier transform in any language I like and use it as I see fit. If, on the other hand, I want to use the one you implemented I have to abide by the license you chose. It's not the algorithm itself what's copyrighted, but the work making use of it. Similarly, there's no copyright on, say, the events that took place during the spanish civil war, but if I write an essay about it I will hold some rights over its distribution and use.

                    Patents are a completely different issue. Just note how I can't implement and distribute an H264 encoder under certain jurisdictions without the risk of patent litigation. The fact that a codec is nothing more than a set of clever mathematical operations to perform a very specific task makes the situation somewhat ridiculous. Can you imagine a fast Fourier transform algorithm protected under a patent? (well, I hope nobody patented anything of the sort)

                    Comment


                    • #60
                      I just found this link in LWN:

                      http://www.nvidia.com/object/tegra-2.html

                      That Tegra 2 thing offers VP8 hardware encoding/decoding.

                      Comment

                      Working...
                      X