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 21st June 2013, 02:26   #21  |  Link
Guest
Guest
 
Join Date: Jan 2002
Posts: 21,922
I have moved the trolling and related discussion to this (closed) thread:

http://forum.doom9.org/showthread.php?t=168116

Please do not troll or feed trolls in this thread because at this point any followups to it here would likely violate rules 3, 4, 11, and/or 16.

Thank you for your understanding.

Last edited by Guest; 21st June 2013 at 02:34.
Guest is offline   Reply With Quote
Old 21st June 2013, 03:59   #22  |  Link
thebombzen
Registered User
 
Join Date: Feb 2011
Posts: 10
Thank you.

Anyway, how much gain can we really expect from CTUs dividing into CUs in practicality, rather than in theory? I found a good description of CTUs/CTBs/CUs/CBs on this PDF from the HHI, if it helps:
http://iphome.hhi.de/wiegand/assets/...erformance.pdf

Will 32x32 and 64x64 CTUs improve compression that much? Or will most of the gain come from the additional encoder flexibility with variable-size non-grid CUs within CTUs? Also, will the additional encoder flexibility allow for encoders which are much, much better than the bitstream format devs expected? x264 is a very good encoder for H.264, better than the H.264 devs expected, but I'm worried it's reached a point where it doesn't have any additional bitstream flexibility to allow it to improve (the x264 devs can give me an opinion on this, I'd be glad to be wrong.)

TL;DR: How much difference does the additional flexibility really make?
thebombzen is offline   Reply With Quote
Old 21st June 2013, 23:04   #23  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 3,343
Quote:
Originally Posted by thebombzen View Post
Thank you.

Anyway, how much gain can we really expect from CTUs dividing into CUs in practicality, rather than in theory? I found a good description of CTUs/CTBs/CUs/CBs on this PDF from the HHI, if it helps:
http://iphome.hhi.de/wiegand/assets/...erformance.pdf
Overall, there's no One Big Thing that makes HEVC so much better. It's the aggregation of lots of little things.

Quote:
Will 32x32 and 64x64 CTUs improve compression that much? Or will most of the gain come from the additional encoder flexibility with variable-size non-grid CUs within CTUs? Also, will the additional encoder flexibility allow for encoders which are much, much better than the bitstream format devs expected? x264 is a very good encoder for H.264, better than the H.264 devs expected, but I'm worried it's reached a point where it doesn't have any additional bitstream flexibility to allow it to improve (the x264 devs can give me an opinion on this, I'd be glad to be wrong.)
Those big CTUs are likely to be quite helpful for 4K/UHD resolutions.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 23rd June 2013, 08:00   #24  |  Link
dapperdan
Registered User
 
Join Date: Aug 2009
Posts: 196
When the trolling got cut out, we also lost a link to a blog comparing hevc and x264, which wasn't particularly interesting in itself but had a link to this recent paper which was:

https://ece.uwaterloo.ca/~z70wang/pu...ons/vpqm13.pdf

The basic gist is that all the standard objective frame by frame analysis methodologies (psnr, ssim etc.) may systematically, and substantially, underrepresent the improvements in HEVC over its predecessor.

Last edited by Guest; 23rd June 2013 at 12:58. Reason: make link hot
dapperdan is offline   Reply With Quote
Old 23rd June 2013, 09:57   #25  |  Link
Bathrone
Curious Beta Tester
 
Join Date: Nov 2004
Posts: 240
This bloke has updated his posts and says that this month he will have a new code release out:

https://code.google.com/p/x265/

Last edited by Guest; 23rd June 2013 at 12:59. Reason: make link hot
Bathrone is offline   Reply With Quote
Old 26th June 2013, 16:09   #26  |  Link
jq963152
Registered User
 
Join Date: Apr 2012
Posts: 691
Can we please have some comments from Dark Shikari and/or akupenguin (or other appropriate x264 developers) regarding if they are already working on an open-source H.265 encoder?
jq963152 is offline   Reply With Quote
Old 26th June 2013, 20:24   #27  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 3,343
Quote:
Originally Posted by jq963152 View Post
Can we please have some comments from Dark Shikari and/or akupenguin (or other appropriate x264 developers) regarding if they are already working on an open-source H.265 encoder?
Do note that if one or both are doing something, it could well be under NDA, and thus they might not be able to respond one way or another.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 26th June 2013, 21:03   #28  |  Link
Dark Shikari
x264 developer
 
Dark Shikari's Avatar
 
Join Date: Sep 2005
Posts: 8,688
There is something in the works, and hopefully you're hear about it soon. Keep in mind that this is going to take years to produce a good encoder if ever, just like with x264. It's not going to be ready to displace H.264 by Christmas.

I also kind of don't feel like engaging trolls, so...
Dark Shikari is offline   Reply With Quote
Old 26th June 2013, 21:53   #29  |  Link
jq963152
Registered User
 
Join Date: Apr 2012
Posts: 691
Thanks for your reply.

Quote:
Originally Posted by Dark Shikari View Post
There is something in the works
Awesome .

Quote:
Originally Posted by Dark Shikari View Post
and hopefully you're hear about it soon.
Would be great.

Quote:
Originally Posted by Dark Shikari View Post
Keep in mind that this is going to take years to produce a good encoder if ever, just like with x264.
So you can't use x264 as a base to speed up development?

Quote:
Originally Posted by Dark Shikari View Post
It's not going to be ready to displace H.264 by Christmas.
Quite unfortunate .

Another question:

Have you already come up with a name? Because it can't be "x265" because that one is already in use?
jq963152 is offline   Reply With Quote
Old 26th June 2013, 22:10   #30  |  Link
Dark Shikari
x264 developer
 
Dark Shikari's Avatar
 
Join Date: Sep 2005
Posts: 8,688
"Using x264 as a base" is not a magic silver bullet that makes developer-years of work go away. It might be called x265; last I recall the domain and trademark had been taken care of.
Dark Shikari is offline   Reply With Quote
Old 26th June 2013, 22:33   #31  |  Link
phate89
Registered User
 
Join Date: Apr 2009
Posts: 153
Quote:
Originally Posted by Dark Shikari View Post
"Using x264 as a base" is not a magic silver bullet that makes developer-years of work go away. It might be called x265; last I recall the domain and trademark had been taken care of.
Sure but years of experience (especially with psychovisual enhancements) with x264 is a better start point right?
Are you planning some in-depth analysis of the standard like you did with vp8 in your diary of x264 dev?
phate89 is offline   Reply With Quote
Old 27th June 2013, 00:56   #32  |  Link
thebombzen
Registered User
 
Join Date: Feb 2011
Posts: 10
Quote:
Originally Posted by phate89 View Post
Sure but years of experience (especially with psychovisual enhancements) with x264 is a better start point right?
Are you planning some in-depth analysis of the standard like you did with vp8 in your diary of x264 dev?
My guess it is will help some but not as much as everyone would like. HEVC is not H.264 so different enhancements might still look good but some other equivalent quality enhancement will compress much better. That's what this kind of encoder research is for
thebombzen is offline   Reply With Quote
Old 27th June 2013, 01:00   #33  |  Link
Dark Shikari
x264 developer
 
Dark Shikari's Avatar
 
Join Date: Sep 2005
Posts: 8,688
Years of experience gives some guidance in experimentation and optimization, but you still have to build the software in the first place and make it fast, clean, and efficient, and that alone is many developer-years of work. Applying smart algorithms and knowledge gleaned from x264 comes after that, and still, it's just guidance; H.265 is very different from H.264.
Dark Shikari is offline   Reply With Quote
Old 27th June 2013, 08:46   #34  |  Link
easyfab
Registered User
 
Join Date: Jan 2002
Posts: 327
I agree It will take years for a good H.265 encoder ( i.e ~50% bitrate for the same H.264 quality )
But I hope the first generation of H.265 encoder will quickly at least be as good as current H.264 encoder in bitrate/speed/quality.
easyfab is offline   Reply With Quote
Old 27th June 2013, 16:27   #35  |  Link
Dark Shikari
x264 developer
 
Dark Shikari's Avatar
 
Join Date: Sep 2005
Posts: 8,688
Keep in mind that it took until 2007-2008 for x264 to consistently beat Xvid (the previous generation), and H.264 came out in 2003. I don't want people to get their hopes too high, or else everyone's going to be disappointed.
Dark Shikari is offline   Reply With Quote
Old 28th June 2013, 05:14   #36  |  Link
Bathrone
Curious Beta Tester
 
Join Date: Nov 2004
Posts: 240
Heres a group with a free HEVC software decoder for windows:

http://xhevc.com/en/hevc/decoder/download.jsp

Divx have a patch out for mkvtoolknix to add their divx HEVC profiles to the toolset. They also reckon they are close to releasing a HEVC decoder and their own encoder.
Bathrone is offline   Reply With Quote
Old 28th June 2013, 07:11   #37  |  Link
JEEB
もこたんインしたお!
 
JEEB's Avatar
 
Join Date: Jan 2008
Location: Finland / Japan
Posts: 514
Quote:
Originally Posted by Bathrone View Post
Heres a group with a free HEVC software decoder for windows:

...
I think these are the guys that kierank caught using libavcodec stuff without properly dealing with it even under the LGPL. I wouldn't give these guys too much publicity here.

Also, I really didn't want to put the same stuff in a yet another thread (we already have HEVC general and Available HEVC/H.265 test encoders threads, as well as some other random ones...), but there is a libavcodec HEVC decoder by smarter that is by now in a pretty good condition. It has been forked by OpenHEVC, and is also developed there (in a somewhat chaotic form).

OpenHEVC is then used by GPAC for their Osmo4 "mp4" player. GPAC happens to have one of the 14496-15 AMD2 (HEVC File Format, aka HEVC-in-"MP4") editors on their team, so they are currently the only ones who can implement it until the last draft is published (the newest draft currently public is from May 2012, and there seem to have been plenty of changes since). Since GPAC is not historically well known for implementing things well/correctly, no-one is copying their implementation, but instead waiting for the last draft to be published to implement it.

Quote:
Originally Posted by Bathrone View Post
Divx have a patch out for mkvtoolknix to add their divx HEVC profiles to the toolset. They also reckon they are close to releasing a HEVC decoder and their own encoder.
They have a patch out for a WIP first draft of HEVC-in-Matroska. You can see the discussion related to it here. I really should have noted that they should have also added the EXPERIMENTAL flag there, since so many people want to grab the extradata format from 14496-15 AMD2, which does not have its current draft released yet. Thus one should NOT think that HEVC-in-Matroska produced with their patched binaries will be compatible with what will end up being specified as the official mapping for HEVC in Matroska. Same warning of course goes for using the GPAC implementation of 14496-15 AMD2, as that can have bugs as well as the draft in general can change with time until the final draft gets published for the final ballot.
__________________
[I'm human, no debug]
JEEB is offline   Reply With Quote
Old 28th June 2013, 09:57   #38  |  Link
iwod
Registered User
 
Join Date: Apr 2002
Posts: 757
Quote:
Originally Posted by Dark Shikari View Post
Keep in mind that it took until 2007-2008 for x264 to consistently beat Xvid (the previous generation), and H.264 came out in 2003. I don't want people to get their hopes too high, or else everyone's going to be disappointed.
Yes, sometimes i wish throwing in money would speeds things up a lot, then we could do a Kickstarter project.
iwod is offline   Reply With Quote
Old 28th June 2013, 15:46   #39  |  Link
Bathrone
Curious Beta Tester
 
Join Date: Nov 2004
Posts: 240
Quote:
Originally Posted by JEEB View Post
I think these are the guys that kierank caught using libavcodec stuff without properly dealing with it even under the LGPL
Thanks for that. If they are abusing open source and it's a proven situation then those abusers should be named and shamed.

The topic of this thread is about HEVC decoders/encoders so its ontopic to discuss various implementations of that. I'm excited mate to hear about the libavcodec decoder and I'm a regular with ffmpeg so thats great
Bathrone is offline   Reply With Quote
Old 30th June 2013, 13:43   #40  |  Link
BadFrame
Registered User
 
Join Date: Jun 2013
Posts: 98
Quote:
Originally Posted by Bathrone View Post
This bloke has updated his posts and says that this month he will have a new code release out:

https://code.google.com/p/x265/
Sounds interesting, there's mentioning of a commercial version, do you know if will this be like x264 where everything is open source but there is an option to purchase a proprietary licence, or will this be two versions where the commercial one is better?

Anyway, great hearing that the x264 devs have something cooking for h265, their track record speaks for itself.
BadFrame 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 20:44.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, vBulletin Solutions Inc.