Announcement

Collapse
No announcement yet.

A Look At Daala's Git Repository, The Lead Developers & Code Count

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

  • Gusar
    replied
    Originally posted by wagaf View Post
    Well, unlike Microsoft, Google controls content distribution through YouTube. Just YouTube using VP9 is already big.
    It is big, but it's still just one site. What about Netflix? They're also extremely big. What about Hulu? Amazon? Broadcast network's or cable channel's own streaming sites? Then, what about disc formats? Microsoft got VC1 into Blu-ray, did Google even try getting VP9 into UHD? The next disc format (if there will even be one)?

    Is Google actively working with any standards organization, be it SMPTE, ITU, IETF? Also, where are competing implementations of encoders? And what about professional video editors or authoring tools, do any of them work with VP9?

    You see, there's quite a bit more than Youtube out there.

    Leave a comment:


  • wagaf
    replied
    Originally posted by Gusar View Post

    Considering not even VC1 made it, even when it did go through a standardization process, what chance does VP9 have of going beyond Youtube? Basically none.
    It's too late for VP9, but if there's gonna be a VP10, Google shouldn't develop it by themselves. They should work on the format with someone like IETF, so that libvpx bugs won't become part of the format, so that there will be a formal specification, so that it won't be a proprietary Google format but a truly open standard.

    Daala is still some time away, VP9 I've described above, the HEVC licensing situation is a mess due to the new patent pool... That means h264 will be with us for quite some time.
    Well, unlike Microsoft, Google controls content distribution through YouTube. Just YouTube using VP9 is already big.

    If anyone wants to build a Free software including some sort of video support, VP9 is a good choice because it's both patent-"free" (unlike VC1 and h264) and well supported by hardware.

    Leave a comment:


  • Gusar
    replied
    Originally posted by xeekei View Post
    All Daala would really need is hardware support from Intel. Then it'd be a slam dunk.
    Guys, slow down. Daala is still deep in the research phase. Not even the creators themselves know how the final Daala bitstream will look like or what technologies it will use.

    It'll take at least a year, probably more, before Daala will become something.

    But here you are, talking about hardware decoders. Let me give you some perspective - The HEVC format was formally released in April 2013. But we're only now, two years later, getting hardware decoders.

    Leave a comment:


  • xeekei
    replied
    All Daala would really need is hardware support from Intel. Then it'd be a slam dunk.

    Leave a comment:


  • profoundWHALE
    replied
    If Daala gets into a standard much like Xiph has with Theora and Ogg/Opus with the HTML5 standard, it's not a matter of if it will be supported, but when.

    Leave a comment:


  • oibaf
    replied
    The long term proposed solution for video is NetVC which should possibly based on Daala:

    Leave a comment:


  • profoundWHALE
    replied
    Originally posted by blubbaer View Post
    VP9 or VP10 (when it will be released) have better chances because Google is pushing it. Some SoC already have hardware acceleration for VP9 and also some SmartTVs already support VP9.
    Google has expressed interest in having Daala for VP10, but it would have to be better than or about equal to what they are working on for VP10/h.266-or-whatever. The other reason for choosing Daala would be to avoid all patents involving h.264/265
    Last edited by profoundWHALE; 07 August 2015, 01:47 PM.

    Leave a comment:


  • Gusar
    replied
    Originally posted by blubbaer View Post
    VP9 or VP10 (when it will be released) have better chances because Google is pushing it. Some SoC already have hardware acceleration for VP9 and also some SmartTVs already support VP9.
    VP9 has its own problems. It's effectively a proprietary format. In the sense that only Google was involved in creating it. In the sense that there's no formal, complete specification, there's just the reference implementation - Google's own libvpx. And, as is quite common in such a case, this leads to things like what is mentioned here - there's a bug in the libvpx loop filter code and this bug must be replicated in every other implementation of the codec. Had the codec gone through a standardization process, such a thing wouldn't happen.

    Also, while Google managed to get device vendors to put hardware decoders in their devices, there's just one provider of VP9 video - Google's own Youtube. And this will highly likely not change. Also, being Google's format, there's no interest by any company to write an independent encoder. Contrast that to the MPEG family of codecs where there's several companies working on competing implementations and writing tools for various different scenarios and workflows.

    The above means we're basically at the mercy of Google and their libvpx team when it comes to an encoder. A team that only fairly recently started working on psychovisual optimizations (even though it's exactly the years long psy-opts tuning that makes x264 such an awesome encoder), a team whose priority doesn't seem to be properly parallelizing the encoder to make it more suitable for home use.

    What Google should've done is something like what Microsoft did when they wanted to get into the HD-DVD and Blu-ray business. Microsoft didn't just release the WMV source code. Instead, they gave the format to SMPTE (Society of Motion Picture & Television Engineers), who cleaned it up, made it suitable for containers other than Microsoft's own (ASF), added what they deemed important back then - interlaced encoding, and wrote a specification. Thus VC1 was born.
    Though not even VC1 made it. While several HD-DVDs and even some Blu-rays are encoded with it (the Serenity Blu-ray I used in my encoder test is VC1), there was only ever one implementation - Microsoft's. And this business didn't work out for Microsoft, at some point they disbanded their media team, and that was that for VC1.

    Considering not even VC1 made it, even when it did go through a standardization process, what chance does VP9 have of going beyond Youtube? Basically none.
    It's too late for VP9, but if there's gonna be a VP10, Google shouldn't develop it by themselves. They should work on the format with someone like IETF, so that libvpx bugs won't become part of the format, so that there will be a formal specification, so that it won't be a proprietary Google format but a truly open standard.

    Daala is still some time away, VP9 I've described above, the HEVC licensing situation is a mess due to the new patent pool... That means h264 will be with us for quite some time.
    Last edited by Gusar; 07 August 2015, 01:48 PM.

    Leave a comment:


  • zxy_thf
    replied
    Originally posted by spinkham View Post
    I'd be skeptical about the chances too, but the content based pricing of the new MPEG h265 patent pool makes it a lot more attractive.
    Now I hope these greedy people will success, which can drive major companies away from patent protected codec standards.

    Leave a comment:


  • CrystalGamma
    replied
    Originally posted by blubbaer View Post
    VP9 or VP10 (when it will be released) have better chances because Google is pushing it. Some SoC already have hardware acceleration for VP9 and also some SmartTVs already support VP9.
    Which is why Daala aims for the generation AFTER VP9.
    Also, remember how well Opus, which was developed by Xiph as well, was received ... even Youtube is using it now!

    Leave a comment:

Working...
X