back to article Web giants gang up to take on MPEG LA, HEVC Advance with royalty-free streaming codec

Some of the largest companies in online media have banded together to battle back against excessive patent licensing fees for streaming video. Amazon, Cisco, Google, Intel, Microsoft, Mozilla, and Netflix have all joined the Alliance for Open Media, a new consortium aimed at developing a new, open source, royalty-free video …

  1. Salts

    Old media meet...

    New media, to you they look like a fu|<ing big meteor; we can all hope :-)

    1. Anonymous Coward
      Anonymous Coward

      Re: Old media meet...

      It will end up same as before. The Old Media will find a shill to go in front of a standards meeting and offer the codec "free" for all to use in a manner which deliberately sabotages the standards process and disarms any such ideas about royalty free. The sole difference between then and now is that Microsoft and Cisco were on the side of Old Media that time.

      Cisco did it at IETF in WebRTC WG with H264 a couple of years back. It will be interesting which company will be used to front the "charitable action" this time.

      By the way, I have some serious doubts about the genuine desire of those two to come up with a free codec. Their previous actions @ IETF and other standards bodies show quite clearly that they are ready to do anything and everything to ensure that MPEG-LA retains its position in the market.

  2. Anonymous Coward
    Anonymous Coward

    A bit late considering Windows 10 already ships with HEVC so whatever they come up with they'll probably have an uphill battle.

    1. tom dial Silver badge

      And yet Microsoft is a member of the Alliance for Open Media that is developing the new codec. Perhaps they also would like to be done with the need to pay royalties for patents that many think ought not to have been issued.

      1. Charles 9

        "And yet Microsoft is a member of the Alliance for Open Media that is developing the new codec. Perhaps they also would like to be done with the need to pay royalties for patents that many think ought not to have been issued."

        And yet Microsoft is part of MPEG-LA last I checked, meaning they get a cut of the proceeds.

        1. Dan Paul

          @Charles 9

          The problem is that if you don't have the codec in the OS, you can't play the video.

          MS operating systems run far too many video management systems and security systems to not have that capability baked in to the OS. Let alone the punters on Youtube.

          They can always change codecs later after the new standard, but they have to support the old one until then or they break compatibility.

    2. Bob H

      It is easy to add codecs to computers, but the bigger question is consumer electronics devices. Add a new codec and will it be supported by phones, tablets, set-top boxes, digital TVs, etc? Computers may be getting good use but the most hours watched remains on embedded devices with fixed ASICs. Everyone got excited some months back when V-Nova uncloaked that they had a new codec which, they claim, is more efficient than anything on the market but until it becomes viable for embedded devices I remain sceptical.

  3. Anonymous Coward
    Anonymous Coward

    Yay... But...

    We have VP9 already - we need less talking shops, more whacking hardware vendors to support existing non-patented codecs...

    It takes:

    - 1 year to develop the codec

    - 1 year to argue about it

    - 1 year to get a software implementation into most browsers

    - 1 year for hardware vendors to get it into their flagship product

    - 1 year for it to be a standard GPU feature

    - 3 years for a consumer refresh cycle

    Vp9 is like half way through that list...

  4. Anonymous Coward
    Paris Hilton

    Ye canna break the laws of physics

    "One major goal is to create something that reduces the amount of bandwidth needed for streaming 4K video, which is still out of reach for most customers but will be a biggie for the likes of Netflix."

    4K is a bit of a nebulous term but in the end if you want to put 4096 x 2160 at say 24 bit colour on a screen, you are going to have to shift rather a lot of data, say 202 Mb (4096 * 2160 * 24 / (1024 * 1024) ~= 202 ). So, your 4K thing is bollocks, cos the above is per frame and there are quite a few frames per second: say 25 (202 * 25 = 5Gbs-1!) So you use funky thingies to compress the stream: Fast Fourier transforms and other fancy mathematical tricks. You also mess with the speed of frames depending on the scene etc etc.

    Anyway, let's say you are in the UK on a really decent connection, say a nominal 80/20Mbs-1 FTTC. That 80 is a maximum and is rather smaller than 5000. Let's go for the median - 40, so 5000/40 = 500/4 (it's worse than that 1000 != 1024) = 125. ie 125 times more data than you can stream on that connection.

    The above is a bit of a joke (and probably gains or drops a zero or two) but when you look at the raw numbers of what is claimed to be offered. it rapidly falls apart.

    1. RichardB

      Re: Ye canna break the laws of physics

      Sure, if you assume that every adjoining pixel is a different colour and changes every frame.

      But usually when I watch tv there are only minimal changes between frames, other than on those somewhat blurry car chases (and even then the whole point of the 'movie' is to not change too much between scenes to fool the eye into thinking it is movement), and cut scenes (which can easily be staggered over a number of frames with minimal interruption to a viewer).

    2. Warm Braw

      Re: Ye canna break the laws of physics

      Well, 1080p is very roughly 200MB/sec, uncompressed, whereas Freeview HD manages to transmit a compressed facsimile of that data at under 8Mb/sec, that's a 200:1 compression rate. So even assuming a similar compression ration to H.264, your 4K video is already down to 25Mb/sec.

      1. This post has been deleted by its author

        1. Warm Braw

          Re: Ye canna break the laws of physics

          "Remember that live bitrate and multi-pass OTT encoding are different."

          That's why I said "even assuming a similar compression ratio" to operational broadcast technology, the situation is way better already than gerdesj was implying. If you don't have to do your compression in real time (across a multiplex of different programme streams), of course you should be able to better than that.

  5. jnffarrell1

    Pity the poor video trolls. Big companies are ganging up on them. Quick call in the SOPS vampires (the undead seeking blue ray forever).

    1. Bob H

      Troll or not?

      To be fair many of these codecs are built on solid research that takes many, many years. I wouldn't call them patent trolls because a patent troll tends to depend on someone else's work and most of the MPEG LA patent holders are genuine researchers. What is good is that some of the more fundamental mathematics which have driven MPEG2 and MPEG4 are expiring in patent so many simple applications basic mathematics can no longer be charged for. If a company spends millions on research they perhaps deserve to get revenue from it?

      1. nijam Silver badge

        Re: Troll or not?

        I'm sure that fundamental mathematics can't be patented. But no doubt some troll pressure group is trying to persuade the USA(1) to rectify that glaring oversight.

        (1) Why the USA? Because (a) they're easily swayed and (b) by virtue of international treaty, the rest of the world seems to be pretty well stuck with their idiotic patent/copyright/... laws.

      2. Tom 13

        Re: more fundamental mathematics

        Even US patent and copyright law (screwy as it is) doesn't permit you to patent fundamental mathematics.

        Therefore MPEG LA are patent trolls.

        1. Anonymous Coward
          Anonymous Coward

          Re: more fundamental mathematics

          But what is their definition of "fundamental"? Is it very basic stuff like 1+1=2 or does it go further than that?

          And furthermore, for those who say virtual stuff can't be patented, they can always implement the design on silicon and say here's an example, fulfilling any "physical design" condition.

          1. tom dial Silver badge

            Re: more fundamental mathematics

            RSA was patented, as was Karmarkar's algorithm for solution of linear programming problems and LZW data compression. Also, of course, the algorithms MPEG LA have been squeezing for years. Some might not like it, and think it should be changed, but the facts are as they are: a cleverly described piece of mathematics can be issued a US patent.

  6. Loyal Commenter Silver badge

    Oh Goody

    Another codec to add to the list that won't be supported by existing devices.

    Obligatory XKCD

  7. Anonymous Coward
    Anonymous Coward

    There will never be a royalty free codec

    Whatever they develop is going to hit patents owned by someone. The whole "HEVC Advance" thing is regarding patents they didn't identify during the standards process, and there are definitely more HEVC related patents out there (along with other patents with owners who think they are part of HEVC and will require the expenditure of millions in legal fees by other parties to prove they are not)

    So Google can develop VP10 or this consortium can develop something from scratch, big deal. Either way, just because they don't hit anything they know is patented doesn't mean anything. If it became widely adopted like VP8/VP9 have not been, patent owners will come out of the woodwork and demand their fair (or more likely unfair) share. They have zero incentive to make themselves known early in the process, they want to wait until it is widely used so it is too late to turn back. Sort of like how HEVC Advance showed up after HEVC had been selected as the standard all cable/satellite companies will be using for 4K, and has been built into just about every one of the tens of millions of 4K TVs that have been sold in the past couple years.

    1. Charles 9

      Re: There will never be a royalty free codec

      But you forget. The consortium has patents of their own (take Google and how they bought the company that developed the VP codec series). Meaning if a submarine patent does emerge, whatever they're using it on is likely to be in conflict with one of their patents, meaning attempting to attack the consortium risks a patent war in court, with the possible result of their patent being invalidated. That's probably one reason MPEG-LA stopped attacking Google over VP8: because Google got patents in the buyout, too, which could potentially snarl AVC.

      1. Anonymous Coward
        Anonymous Coward

        Re: There will never be a royalty free codec

        I wasn't suggesting patent battles might occur with MPEG-LA or their members like Apple and Sony. I'm talking about patent trolls or formerly great companies that are on their last legs and are essentially becoming a patent troll that make no products and thus have no need to license anyone else's stuff leaving no way to retaliate against them.

        They have nothing to lose by suing and everything to gain by waiting a few years before filing that suit to see if stuff like VP10, Cisco's codec etc. gain some market traction before striking.

        1. Anonymous Coward
          Anonymous Coward

          Re: There will never be a royalty free codec

          "They have nothing to lose by suing and everything to gain by waiting a few years before filing that suit to see if stuff like VP10, Cisco's codec etc. gain some market traction before striking."

          Even if they're found to be at fault and nailed with the attorney's fees for both sides because of it? Under such a cloud, they'd probably be more inclined to take a buyout.

          1. Anonymous Coward
            Anonymous Coward

            Re: There will never be a royalty free codec

            That's pretty rare (talking about US law here) since so long as they believe in good faith that their patent is applicable they may lose the case, but rarely get slapped with the defendant's attorney fees. Even if they are, that's not really a problem, because patent trolls are shell companies without assets. When they strike it rich on a valuable patent that's earning big revenue they split that off into a different corporation so the assets are untouchable in cases regarding other patents.

            At any rate, I'm thinking about situations where the trolls have patents that really are applicable. It is impossible to design any codec without using tons of stuff that has been done, and patented, previously. When companies that originally developed the technology go under, their patents are often picked up by trolls if they can get them cheap (or have lawyers smart enough to recognize they have real value and are worth spending real money on) so they aren't necessarily using bad patents. Well...they are software/algorithm patents, which IMHO shouldn't be patentable, but so long as they are those are the rules of the game.

  8. Zmodem

    use xvid, its mpeg-4 iso and nothing todo with mpeg-4 part 2, and is better then your hevc and mpeg-4 advanced, and better then divx, and 4 times faster to encode

    you can download my settings on http://forum.digital-digest.com/f75/xvid-settings-94010.html

    1. Anonymous Coward
      Anonymous Coward

      There's a reason it's called xvid. It's based on DivX (eman sdrawkcab), which is based on MPEG-4 Part 2 (AVC is Part 10). MPEG-4 ISO is the entire spec, including parts 2 and 10.

      HEVC is also known as H.265 and MPEG-H Part 2 (MPEG-H ISO was developed after MPEG-4 ISO) so was built on experience with MPEG-4 and frequently does better than any MPEG-4 implementation under the same bandwidth constraints.

      1. Zmodem

        divx has never been open source, when it comes down to the crunch and you want to encode festivals like

        https://www.youtube.com/watch?v=7T0NrGr8d-c

        xvid has always been the best for picture quality and file size, with no blocks or rubbish gradients when smoke machines kick in, and also seek time when you want to fastforward or rewind

        xvid is already in use on more dvd, blu-ray players and camera`s

        1. Charles 9

          Zmodem, xvid is supported on those devices because MPEG-4 Part 2 (which the DivX codec used) is supported. xvid is nothing more or less than an open-source version of MPEG-4 part 2. It even supports the same DivX encoder profiles.

          If you honestly think xvid is all that, here's a way to end the debate: take a high-motion high-def event (a good example would be an auto race) and encode it both with xvid and with x264 (an open-source implementation of AVC), using the same bandwidth constraints. There's a reason Part 2 was superceded by Part 10 and now is being challenged by HEVC.

          PS. Last I checked, YouTube doesn't use Part 2, meaning that clip you linked was using AVC (Part 10).

          PSS. What's the problem with x264, for that matter? It's open source just like xvid.

          1. Zmodem

            mpeg-4 iso is just a container, and everything else is custom, encoding a hi speed race means nothing, any encoder can handle FPS, frames are just compressed images, the compression takes away the detail, encoding festival mainstages are where most codecs fail except for xvid

            xvid and divx are seperate codecs, most programs use profile parsing to import the basic values

            if anything 3ivx is using an old xvid source code

            festival stages are just your sci-fi movie`s which looks rubbish on DVD, and still abit rubbish on blu-ray

            1. Anonymous Coward
              Anonymous Coward

              Don't believe me? Read the spec for yourself.

              Part 2 is described as "A compression format for visual data". That's NOT a container. The container format is describe in Parts 1, 12, and 14. The AAC audo format common in MP4 is Part 3. And as said before, AVC is Part 10.

              And encoding a high-speed race means encoding high degrees of motion. MPEG and its children all try to optimize by way of motion estimation, so when you have lots of motion, especially from one frame to the next (which you'd have in an auto race, where the camera has to pan quickly to keep up with a car traveling past at 200+mph), you trip this up. Not only that, races are usually live events, so you have to use a realtime encoder which places additional limitations on your ability to capture the action properly.

              PS. If mainstages cause codecs to fail, explain the YouTube clip, which is AVC-encoded (YouTube does not use xvid because it either uses AVC—for standards reasons—or VP8/WebM—because Google owns YouTube). If xvid is truly superior, even against x264 and x265—all of which are open-source—then show the proof with a side-by-side.

              PSS. If you're saying stages are like sci-fi movies, then perhaps your beef is that they have trouble with low contrast (very dark areas) which tends to show blockiness. Weaker codecs are actually masking this by making it look like a washed-out monotone while newer codecs at least endeavor to preserve what little details there are.

              1. Zmodem

                if xvid can do gradients better then other mpeg-4 codecs with lower bitrates and smaller filesizes then xvid would be using its own compression

                this scene in predator 2 looks terrible on dvd, I have`nt seen it on blu-ray

                http://www.avpgalaxy.net/wordpress/wp-content/uploads/2011/06/p2-ship07.jpg

                most codecs will end up with 256 color gradient, instead of 16 million

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon