Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Video Encoding > High Efficiency Video Coding (HEVC)

Reply
 
Thread Tools Search this Thread Display Modes
Old 11th June 2013, 05:30   #1  |  Link
thebombzen
Registered User
 
Join Date: Feb 2011
Posts: 10
H.265/HEVC specification released, any decoders/encoders in sight?

Hi, According to Wikipedia, the H.265/HEVC spec was released on 7 June 2013, which I found for free on ITU-T's website:

http://www.itu.int/rec/T-REC-H.265-201304-I/en

Now that HEVC has been officially finalized and released, corporations (ATEME, DivX, maybe not Google) will be jumping up and down to quickly put forth their own HEVC decoders and encoders. However, as history has shown, open-source software often beats proprietary software. Look at x264, OPUS, nothing has wider support than FFmpeg, and Linux servers are an industry standard.

While on Wikipedia it says that many companies are now putting their feet forward in the HEVC market, does anyone have an estimation for free and open-source (BSD, (L)GPL, etc.) decoders or encoders? Note that X265 doesn't count, as it's most likely a "lite" version:

Quote:
Originally Posted by X265's Google Code Site
Currently, I working on my commercial version, it is cooperation with a China university, we improvement the compress performance and try more advanced feature (many-cores, taskpool, GPU, etc).
Also, is this move going to kill x264? It seems that does happen for software, for example, the Blender3D Cycles rendering engine effectively killed Luxrender. Codecs for inferior formats such as LAME still live but notice that LAME's last stable release was in 2011 (2 years ago) and probably still exists only because there's no good free as in freedom encoder for AAC (FFmpeg's is terrible, so is vo-aacenc, FAAC and fdk-aac are nonredistributable and Nero's and Apple's are not open-source, etc.).
thebombzen is offline   Reply With Quote
Old 11th June 2013, 14:57   #2  |  Link
qyot27
...?
 
qyot27's Avatar
 
Join Date: Nov 2005
Location: Florida
Posts: 1,062
I can't comment on the rest of it, but
Quote:
Originally Posted by thebombzen View Post
Also, is this move going to kill x264? It seems that does happen for software, for example, the Blender3D Cycles rendering engine effectively killed Luxrender.
IMO, probably not. One of the reasons that x264 eclipsed Xvid as quickly as it did (which still took several years; Xvid's apex period was ca. 2003-2005, with a slow declining drop-off for a couple more years after that) is because MPEG-4 ASP was never as deeply entrenched as MPEG-2, which is still used - most obviously for DVD, but also as an option on (mostly older) Blu-ray discs, and it's still widely used for DTV broadcast (although H.264 is in the process of replacing it for this).

H.264 is arguably even more ubiquitous than MPEG-2 ever hoped to be, so I'd wager it'll still take several years for HEVC to become the dominant format, and then you also have to factor in the amount of time it will take for those in the FOSS community to implement and optimize an encoder for it (that is, if development of rival formats like VP9 or Daala don't throw a wrench into the works), while it still has to compete with x264 being as long-mature as it is.

In perspective, H.264's specs arrived in early 2003, and hdot264 appeared sometime afterward. x264 appeared some time in (late?) 2004, didn't start getting a lot of attention until 2005, and H.264 as a format didn't start dominating over MPEG-4 ASP or SP until 2007 or 2008 in tangible online distribution*. The almost wholesale transition of streaming media to H.264 from the variety of formats they were using before didn't really start to coalesce until what, 2009 or 2010? I'm not really sure when MPEG-2 and VC-1 use on Blu-ray started to drop off, but it was still somewhere in there.

*take Apple and iTunes as examples: many/most of the trailers and music videos available through them prior to formally starting the iTunes Video Store were in either MPEG-4 SP (music videos) or Sorenson 3 (trailers). The Video Store explicitly moved its offered content to H.264, and it was around the same time that the trailers site both mostly moved to H.264 and started distributing 720p and 1080p versions. That's just one example (and probably the only one that matters, at least legally).

There are of course going to be early adopters that like the bleeding edge, but you've got to contend with a lot more standalones and mobile devices that support H.264 and not HEVC, which will impact its rate of adoption in the short term. Active development on Xvid didn't start tapering off until it was in its twilight years, so unless there's just this tidal wave of adoption for HEVC, I'd say that H.264 as a format - and x264 as its premier encoder - still have at least a good 5 years before HEVC's rising level of dominance begins to seriously impact it, at which time the use case will shift more to legacy support for things like disc formats or using H.264 as the 'plays on almost anything and/or not-so-powerful machines' solution that MPEG-4 ASP still manages to hold onto. The 5-year point also is likely to mirror the shift to UHDTV or a new generation of Blu-ray, where HEVC will naturally have broad use.
qyot27 is offline   Reply With Quote
Old 11th June 2013, 22:33   #3  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 2,951
Quote:
Originally Posted by qyot27 View Post
IMO, probably not. One of the reasons that x264 eclipsed Xvid as quickly as it did (which still took several years; Xvid's apex period was ca. 2003-2005, with a slow declining drop-off for a couple more years after that) is because MPEG-4 ASP was never as deeply entrenched as MPEG-2, which is still used - most obviously for DVD, but also as an option on (mostly older) Blu-ray discs, and it's still widely used for DTV broadcast (although H.264 is in the process of replacing it for this).
Plus cable, which is almost entirely MPEG-2 yet. I doubt we've hit the crossover point where even 50% of eyeball-hours are H.264 instead of MPEG-2.

Quote:
H.264 is arguably even more ubiquitous than MPEG-2 ever hoped to be, so I'd wager it'll still take several years for HEVC to become the dominant format
MPEG-2 was the first widely used video codec. Since we're now in a mode of multiple coexisting codec generations, I doubt anything will ever have the market share that MPEG-2 had in, say, 2004.

Quote:
and then you also have to factor in the amount of time it will take for those in the FOSS community to implement and optimize an encoder for it (that is, if development of rival formats like VP9 or Daala don't throw a wrench into the works), while it still has to compete with x264 being as long-mature as it is.
I don't see VP9 or Daala competing with resources with HEVC. VP9 is almost entirely a Google-developed effort, and Daala is a long way from having a locked-down bitstream spec. FOSS developers looking for a mainstream audience are going to be choosing between H.264 and HEVC for the most part.

Quote:
In perspective, H.264's specs arrived in early 2003, and hdot264 appeared sometime afterward. x264 appeared some time in (late?) 2004, didn't start getting a lot of attention until 2005, and H.264 as a format didn't start dominating over MPEG-4 ASP or SP until 2007 or 2008 in tangible online distribution*. The almost wholesale transition of streaming media to H.264 from the variety of formats they were using before didn't really start to coalesce until what, 2009 or 2010? I'm not really sure when MPEG-2 and VC-1 use on Blu-ray started to drop off, but it was still somewhere in there.

*take Apple and iTunes as examples: many/most of the trailers and music videos available through them prior to formally starting the iTunes Video Store were in either MPEG-4 SP (music videos) or Sorenson 3 (trailers). The Video Store explicitly moved its offered content to H.264, and it was around the same time that the trailers site both mostly moved to H.264 and started distributing 720p and 1080p versions. That's just one example (and probably the only one that matters, at least legally).
I'd say that VC-1/WMV was the dominant legal online format 2004-2008. That's what Amazon Unbox, Zune, Xbox Video, Netflix, CinemaNow, and most of the other early online video companies other than Apple were using. WMV had the only widely available Hollywood-approved DRM and hardware-accelerated decode. Apple was rather unique, as they had their own DRM that didn't even try to be interoperable and didn't have HW accelerated decode for anything for a long time.

Quote:
There are of course going to be early adopters that like the bleeding edge, but you've got to contend with a lot more standalones and mobile devices that support H.264 and not HEVC, which will impact its rate of adoption in the short term. Active development on Xvid didn't start tapering off until it was in its twilight years, so unless there's just this tidal wave of adoption for HEVC, I'd say that H.264 as a format - and x264 as its premier encoder - still have at least a good 5 years before HEVC's rising level of dominance begins to seriously impact it, at which time the use case will shift more to legacy support for things like disc formats or using H.264 as the 'plays on almost anything and/or not-so-powerful machines' solution that MPEG-4 ASP still manages to hold onto. The 5-year point also is likely to mirror the shift to UHDTV or a new generation of Blu-ray, where HEVC will naturally have broad use.
Blu-ray 4K is still going to be H.264, for example. The rapid rate of mobile device turnover suggests H.264-only devices should be in the minority by end of 2016. But living room devices live a lot longer; <=2013 TVs, STBs, game consoles, etcetera will be in use way past 2016.

I can imagine HEVC starting to see practical use in 2014 and account for a majority of IP streams in 2016. I don't see H.264 being dominant in 2018 overall, but it will certainly still be widely used for popular devices and formats. Heck, MPEG-2 will still be significant in 2018. It'll be a heterogeneous era. But given the dominance of IP delivery by then, I don't see a variety of codecs as being a major problem.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 12th June 2013, 01:41   #4  |  Link
edison
Registered User
 
Join Date: Dec 2005
Posts: 95
all the next gen gpu(for example:nvidia maxwell) will include a h.265 decoder/encoder.
edison is offline   Reply With Quote
Old 12th June 2013, 03:42   #5  |  Link
wanezhiling
Registered User
 
Join Date: Apr 2011
Posts: 1,169
Quote:
Originally Posted by edison View Post
all the next gen gpu(for example:nvidia maxwell) will include a h.265 decoder/encoder.
btw where did you get the info?
wanezhiling is offline   Reply With Quote
Old 12th June 2013, 07:17   #6  |  Link
HerpaDerp
Registered User
 
Join Date: Apr 2011
Posts: 13
Quote:
Blu-ray 4K is still going to be H.264
Are you talking about "mastered in 4k" or some futuristic Blu-ray product?

I can't imagine why HEVC wouldn't be used for 4K / 8K footage. Variable block sizes used in HEVC would mean 4K and 8K footage would look vastly superior when compared to H.264.

Can somebody shed some light on this?
HerpaDerp is offline   Reply With Quote
Old 12th June 2013, 14:15   #7  |  Link
iwod
Registered User
 
Join Date: Apr 2002
Posts: 753
I wonder why x264 folks hasn't shown any interest in HEVC. Are they going to making HEVC leverage some of their work in x264?
iwod is offline   Reply With Quote
Old 12th June 2013, 14:50   #8  |  Link
sneaker_ger
Registered User
 
Join Date: Dec 2002
Posts: 5,478
Maybe they just haven't made their interest public?

Also, even if they are working on an h265 implementation in the background, you can't be sure it will be free and/or open-source. The current main dev joined the existing open-source x264 project as a student IIRC, so he might now want to make more money with his experience. Of course getting an x264-grade implementation that is available through multiple licenses (one being open/free) would be awesome, but that's not for us to decide.
sneaker_ger is offline   Reply With Quote
Old 12th June 2013, 15:19   #9  |  Link
thebombzen
Registered User
 
Join Date: Feb 2011
Posts: 10
Variable block sizes and blocks up to 64x64 (I think, correct me if I'm wrong) and other features give HEVC way more flexibility than H.264. If x264's methods are ported to HEVC then it won't be a state-of-the-art encoder because HEVC gives so much more freedom than H.264.
thebombzen is offline   Reply With Quote
Old 12th June 2013, 18:46   #10  |  Link
mandarinka
Registered User
 
mandarinka's Avatar
 
Join Date: Jan 2007
Posts: 734
It seems that Mainconcept SDK with HEVC encoding support is also out.
mandarinka is offline   Reply With Quote
Old 13th June 2013, 00:33   #11  |  Link
thebombzen
Registered User
 
Join Date: Feb 2011
Posts: 10
Quote:
Originally Posted by qyot27 View Post
I can't comment on the rest of it, but
x264 appeared some time in (late?) 2004, didn't start getting a lot of attention until 2005
That's about 2-ish years since H.264 was released, so should we expect about two years for a good free & open-source HEVC encoder, if at all? Sooner?
thebombzen is offline   Reply With Quote
Old 13th June 2013, 03:03   #12  |  Link
paradoxical
Guest
 
Posts: n/a
Quote:
Originally Posted by thebombzen View Post
While on Wikipedia it says that many companies are now putting their feet forward in the HEVC market, does anyone have an estimation for free and open-source (BSD, (L)GPL, etc.) decoders or encoders?
You mean other than HM which is both free and open source?
  Reply With Quote
Old 14th June 2013, 00:28   #13  |  Link
jq963152
Registered User
 
Join Date: Apr 2012
Posts: 691
Quote:
Originally Posted by benwaggoner View Post
Blu-ray 4K is still going to be H.264, for example.
Source?

Quote:
Originally Posted by benwaggoner View Post
The rapid rate of mobile device turnover suggests H.264-only devices should be in the minority by end of 2016. But living room devices live a lot longer; <=2013 TVs, STBs, game consoles, etcetera will be in use way past 2016.
Does anyone here know if the recently announced PlayStation 4 and/or Xbox One will be able to decode H.265?
jq963152 is offline   Reply With Quote
Old 15th June 2013, 03:21   #14  |  Link
thebombzen
Registered User
 
Join Date: Feb 2011
Posts: 10
Quote:
Originally Posted by paradoxical View Post
You mean other than HM which is both free and open source?
HM is from the Fraunhofer institute IIRC (I downloaded mine from hhi.fraunhofer.de), and I don't expect Fraunhofer products to remain free as in freedom, at least for binaries. I'm not a license expert so if anyone can correct me on this topic then please do, but fdk-aac binaries are nonredistributable and Fraunhofer even went so far as to try to require distributers of MP3-format files to pay royalties. I'm not sure exactly if HM binaries are redistributable (I'd be thrilled if HM was as free as x264) and when I said "free and open-source" I mean as free as it gets as in x264, FFmpeg, or the Linux Kernel. If HM really is this free, then that's a bonus but to the best of my knowledge this is not the case.
thebombzen is offline   Reply With Quote
Old 15th June 2013, 03:43   #15  |  Link
Dark Shikari
x264 developer
 
Dark Shikari's Avatar
 
Join Date: Sep 2005
Posts: 8,689
HM is free as in freedom, as far as I know (patents aside); more practically though, it's about 3-4 orders of magnitude too slow to be useful and missing a lot of user-facing features.
Dark Shikari is offline   Reply With Quote
Old 15th June 2013, 05:29   #16  |  Link
thebombzen
Registered User
 
Join Date: Feb 2011
Posts: 10
Quote:
Originally Posted by Dark Shikari View Post
HM is free as in freedom, as far as I know (patents aside); more practically though, it's about 3-4 orders of magnitude too slow to be useful and missing a lot of user-facing features.
HEVC was originally supposed to be only around 1/2 to 3 times more computationally complex than H.264, which begs the question of why HM is so slow compared to H.264 implementations. Even though Modern H.264 implementations have had 10 years, I've seen x264 encode literally 100 times faster than HM, which is a far cry from the three-times-the-complexity target. It doesn't make sense that it's this slow, even though it's so new.

Quote:
Originally Posted by The Completely Reliable Experts at Wikipedia
The preliminary requirements for NGVC was the capability to have a bit rate reduction of 50% at the same subjective image quality compared to the H.264/MPEG-4 AVC High profile and computational complexity ranging from 1/2 to 3 times that of the High profile. NGVC would be able to provide 25% bit rate reduction along with 50% reduction in complexity at the same perceived video quality as the High profile, or to provide greater bit rate reduction with somewhat higher complexity.
thebombzen is offline   Reply With Quote
Old 15th June 2013, 08:38   #17  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 9,793
HM is a reference encoder, its not optimized much at all. Comparing it to x264 is rather pointless, if you want you can compare it to JM (the H264 reference encoder), but even that may not be a useful comparison, depending on implementation details.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders

Last edited by nevcairiel; 15th June 2013 at 08:41.
nevcairiel is offline   Reply With Quote
Old 15th June 2013, 14:15   #18  |  Link
qyot27
...?
 
qyot27's Avatar
 
Join Date: Nov 2005
Location: Florida
Posts: 1,062
Quote:
Originally Posted by Dark Shikari View Post
HM is free as in freedom, as far as I know (patents aside)
Specifically, it's under the 3-clause BSD. Right from the code itself:
http://hevc.kw.bbc.co.uk/svn/jctvc-a124/trunk/COPYING

Quote:
Originally Posted by nevcairiel
HM is a reference encoder, its not optimized much at all. Comparing it to x264 is rather pointless, if you want you can compare it to JM (the H264 reference encoder), but even that may not be a useful comparison, depending on implementation details.
Have there been any comparisons between the performance of HM vs. JM? Or is that just too masochistic to think about?

I'd be surprised if HM is actually faster, but if that's true, the development model it has may have affected that. Considering that it took upwards of 15 minutes to generate 360-some-frame, 848x480, 14-bit samples with JM, and that was on a Core i5 and only after breaking the video up into four separate pieces so it could be run in parallel. At that kind of performance level, describing the speed as 'glacial' counts as a bit of a compliment.
qyot27 is offline   Reply With Quote
Old 19th June 2013, 09:01   #19  |  Link
pieter3d
Registered User
 
Join Date: Jan 2013
Location: Santa Clara CA
Posts: 118
Quote:
Originally Posted by thebombzen View Post
HEVC was originally supposed to be only around 1/2 to 3 times more computationally complex than H.264, which begs the question of why HM is so slow compared to H.264 implementations. Even though Modern H.264 implementations have had 10 years, I've seen x264 encode literally 100 times faster than HM, which is a far cry from the three-times-the-complexity target. It doesn't make sense that it's this slow, even though it's so new.
HM is slow to encode, but a fast HEVC decoder can certainly be written. I've seen a pure-software decoder smoothly playing 1080p30 on an iPhone5.
pieter3d is offline   Reply With Quote
Old 19th June 2013, 20:08   #20  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 2,951
Quote:
Originally Posted by avih View Post
Guys/gals, the discussion is a good one, but there's no need to get personal. Please keep it on topic. Thanks.
Yeah.

All, bear in mind x264 devs have contributed x264 technologies and techniques to x262, xvid, and xvp8. In my experience with them, including DS, they are much more pragmatic than religious, and care more about making ever-better looking video than about any particular bitstream or feature.

I would be startled if at least some core x264 contributors weren't already making plans around HEVC. And lots of the good things about x264 are quite applicable to HEVC. A lot could be done just by swapping out parts of the reference encoder with adapted x264 bits. It wouldn't be a fully optimized encoder, but it certainly wouldn't take years to have an encoder that reliably gives better quality than x264 at non-glacial encoding speeds.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Reply

Tags
decoder, encoder, h.265, hevc, open-source

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 04:45.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.