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 > New and alternative video codecs

Reply
 
Thread Tools Search this Thread Display Modes
Old 21st February 2019, 23:16   #1481  |  Link
TD-Linux
Registered User
 
Join Date: Aug 2015
Posts: 34
Quote:
Originally Posted by benwaggoner View Post
I worry that AV1's interframe CABAC dependencies will impair random access enough to make the practical maximum GOP duration a lot smaller. Some of that could probably be addressed via encoder tweaks, at the loss of a little efficiency.
You don't need to worry - AV1 probability dependencies can only come from one of the reference frames, so it doesn't place any additional impairment on seekability.

(also note that CABAC is a misnomer, as it's not binary. The spec doesn't give it an acronym, but dav1d uses MSAC).
TD-Linux is offline   Reply With Quote
Old 22nd February 2019, 03:01   #1482  |  Link
Beelzebubu
Registered User
 
Join Date: Feb 2003
Location: New York, NY (USA)
Posts: 66
Quote:
Originally Posted by TD-Linux View Post
You don't need to worry - AV1 probability dependencies can only come from one of the reference frames, so it doesn't place any additional impairment on seekability.
I think his concern that a decoder (or a stupid player, which is 99.9% of them) don't know this. A good container format (like mp4) can represent the reference structure in its atoms, and then a good decoder + good container + good encoder can do the right thing. But if any one of them fails, you'll have a worse seeking experience if you want to do frame-exact user experience. It's up to all devs to make sure that doesn't happen, and like I said, this is multi-factorial so it's easy to forget and screw up.
Beelzebubu is offline   Reply With Quote
Old 22nd February 2019, 21:41   #1483  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 2,984
Quote:
Originally Posted by Beelzebubu View Post
I think his concern that a decoder (or a stupid player, which is 99.9% of them) don't know this. A good container format (like mp4) can represent the reference structure in its atoms, and then a good decoder + good container + good encoder can do the right thing. But if any one of them fails, you'll have a worse seeking experience if you want to do frame-exact user experience. It's up to all devs to make sure that doesn't happen, and like I said, this is multi-factorial so it's easy to forget and screw up.
Yeah, the goal is for the decoder to determine the minimum sequence of frames required to decode a particular frame. With a IbbbBbbbPbbbBbbbPbbbbBbbbI kind of structure, decoding the last "b" frame in an Open GOP should require just six frames (IPPBIb) in decode order typically. But that requires the reference list because sometimes a b could reference two B frames back and things like that. With multiple reference frames it's impossible to reliably know the hierarchy without knowing what each frame references.

And there are patterns that can be spec-legal but that existing encoders might not do. And then better encoders add those to improve quality.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 22nd February 2019, 22:37   #1484  |  Link
TD-Linux
Registered User
 
Join Date: Aug 2015
Posts: 34
Quote:
Originally Posted by benwaggoner View Post
Yeah, the goal is for the decoder to determine the minimum sequence of frames required to decode a particular frame. With a IbbbBbbbPbbbBbbbPbbbbBbbbI kind of structure, decoding the last "b" frame in an Open GOP should require just six frames (IPPBIb) in decode order typically. But that requires the reference list because sometimes a b could reference two B frames back and things like that. With multiple reference frames it's impossible to reliably know the hierarchy without knowing what each frame references.
Yeah, if you want to do that you'll need a reference list parser. But that's been true for a very long time - even x264 produces streams that require you to do this.

But regardless, whatever structure you pick, the CDFs always follow that same structure, so they are "free" from a seekability point of view.

Last edited by TD-Linux; 22nd February 2019 at 22:42.
TD-Linux is offline   Reply With Quote
Old 23rd February 2019, 00:31   #1485  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 2,984
Quote:
Originally Posted by TD-Linux View Post
Yeah, if you want to do that you'll need a reference list parser. But that's been true for a very long time - even x264 produces streams that require you to do this.

But regardless, whatever structure you pick, the CDFs always follow that same structure, so they are "free" from a seekability point of view.
That's good news. I had heard suggestions otherwise.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 24th February 2019, 23:29   #1486  |  Link
sneaker_ger
Registered User
 
Join Date: Dec 2002
Posts: 5,494
Seems like dav1d is now like 40% faster than libaom on my i5-2500K (no AVX2). And there's a pull request for additional SSSE3 optimizations with like another 50% speedup.

https://code.videolan.org/videolan/d...599#note_29705
sneaker_ger is offline   Reply With Quote
Old 25th February 2019, 00:06   #1487  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 9,837
Once the CDEF patch is merged, it should definitely beat AOM in nearly all situations, in 8-bit decoding anyway. 10-bit/12-bit hasn't been worked on much yet, performance wise.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is offline   Reply With Quote
Old 26th February 2019, 15:32   #1488  |  Link
sneaker_ger
Registered User
 
Join Date: Dec 2002
Posts: 5,494
CDEF patch is merged now.
sneaker_ger is offline   Reply With Quote
Old 26th February 2019, 15:53   #1489  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 9,837
Here is an up-to-date performance comparison of aomdec vs dav1d (I believe some of the tests are still running as I write this)
https://docs.google.com/spreadsheets...gid=1835354908

Single-Threading, it still loses in some cases on SSE 4.1 CPUs (aom implements mostly SSE4.1 for pre-AVX), but Multi-Threading it makes up for that.
One missing part is prep_8tap, which can have a decent impact.

Their goal for releasing dav1d 0.2.0 is to be faster then aomdec in 8-bit in all targeted scenarios.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders

Last edited by nevcairiel; 26th February 2019 at 16:04.
nevcairiel is offline   Reply With Quote
Old 26th February 2019, 17:03   #1490  |  Link
sneaker_ger
Registered User
 
Join Date: Dec 2002
Posts: 5,494
There are some user benchmarks on an iPhone with crazy results.


1080p with 46 to 74 fps. Years ago we told people they would never be able to watch new codecs on a phone if it didn't have ASIC for the codec ...
sneaker_ger is offline   Reply With Quote
Old 26th February 2019, 17:27   #1491  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 9,837
Quote:
Originally Posted by sneaker_ger View Post
Years ago we told people they would never be able to watch new codecs on a phone if it didn't have ASIC for the codec ...
They really still shouldn't want to/have to, since it'll make the phone hot and the battery empty.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is offline   Reply With Quote
Old 26th February 2019, 18:00   #1492  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 2,984
Alliance for Open Media codecs

Quote:
Originally Posted by sneaker_ger View Post
There are some user benchmarks on an iPhone with crazy results.





1080p with 46 to 74 fps. Years ago we told people they would never be able to watch new codecs on a phone if it didn't have ASIC for the codec ...
Fun way to amaze yourself - figure out how long ago you first got a computer that was more powerful than your current phone (benchmark/metric of your choice). A modern iPhone has lots of fast cores with SIMD instructions and a pretty darn powerful GPU. It would smoke any hot gaming rig of 10 years ago, and a typical laptop of 5 years ago.

Generally each new codec generation aims to be more than twice as complex to decode as the prior generation. With Moores Law gains, computing devices get a LOT more than 2x faster in that time.

The bigger challenge with software codecs is getting them integrated into hardware DRM required to play premium content above 480p.

Its exciting to see these perf gains with AV1. Hopefully well get to a reasonably done version of dav1d later this year so we can ballpark decoder requirements for AV1 versus other codecs. Im particularly interested in how much parallelism is possible. VP9 was nearly single-threaded, which was problematic

A big question for AV1s viability is how many extra transistors (and thus how much extra die size and SoC cost) full HW decode will take. Ive not heard any details from anyone who has taped out an implementation yet. Anyone else?
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book

Last edited by benwaggoner; 26th February 2019 at 18:08.
benwaggoner is offline   Reply With Quote
Old 26th February 2019, 18:16   #1493  |  Link
NikosD
Registered User
 
Join Date: Aug 2010
Location: Athens, Greece
Posts: 2,708
Quote:
Originally Posted by nevcairiel View Post
Single-Threading, it still loses in some cases on SSE 4.1 CPUs (aom implements mostly SSE4.1 for pre-AVX), but Multi-Threading it makes up for that.
Unfortunately, in single-threading mode, dav1d looses even using SSSE3 in Dua Lipa clip.

It needs further optimization for pre-AVX2 SIMD, but the multi-threading performance especially for AVX2 is impressive.
__________________
Win 10 x64 (18363.476) - Core i3-9100F - nVidia 1660 (441.41)
HEVC decoding benchmarks
H.264 DXVA Benchmarks for all
NikosD is online now   Reply With Quote
Old 26th February 2019, 20:10   #1494  |  Link
EwoutH
Registered User
 
Join Date: Feb 2019
Location: Delft, Netherlands
Posts: 15
@nevcairiel You're fast

That bench included !604 (prep_8tap), so this is most likely the performance as it's going to look like at 0.2.0 release. Not all prep_8tap functions are converted to ssse3 yet, so there is potentially another 5-12% speedup possible for ssse3, highly depending on content.
EwoutH is offline   Reply With Quote
Old 28th February 2019, 07:32   #1495  |  Link
GTPVHD
Registered User
 
Join Date: Mar 2008
Posts: 246
https://aomediacodec.github.io/av1-avif/

v1.0.0, 19 February 2019 but page still says it is a draft.
GTPVHD is offline   Reply With Quote
Old 28th February 2019, 08:34   #1496  |  Link
iwod
Registered User
 
Join Date: Apr 2002
Posts: 753
Quote:
Originally Posted by GTPVHD View Post
https://aomediacodec.github.io/av1-avif/

v1.0.0, 19 February 2019 but page still says it is a draft.
That is the Image format AVIF you are looking at. Not sure if that is what you want.
iwod is offline   Reply With Quote
Old 28th February 2019, 10:45   #1497  |  Link
Q3CPMA
Registered User
 
Q3CPMA's Avatar
 
Join Date: Jun 2015
Posts: 25
Personally, I find AVIF a lot more interesting than AV1 seeing as JPEG is technologically much more obsolete than AVC (if you don't consider 4K, of course). I just hope they don't force chroma subsampling like on webp.
Q3CPMA is offline   Reply With Quote
Old 28th February 2019, 13:01   #1498  |  Link
hajj_3
Registered User
 
Join Date: Mar 2004
Posts: 905
Quote:
Originally Posted by Q3CPMA View Post
Personally, I find AVIF a lot more interesting than AV1 seeing as JPEG is technologically much more obsolete than AVC (if you don't consider 4K, of course). I just hope they don't force chroma subsampling like on webp.
Jpeg XL will be ratified later this year, that should be much nicer than AVIF.
hajj_3 is offline   Reply With Quote
Old 28th February 2019, 22:05   #1499  |  Link
Rumbah
Registered User
 
Join Date: Mar 2003
Posts: 466
Quote:
Originally Posted by hajj_3 View Post
Jpeg XL will be ratified later this year, that should be much nicer than AVIF.
Will it be royalty free?
__________________
x264 full help - x264 --fullhelp r2345
Cuttermaran HCEnc provider - Support for HCEnc in Cuttermaran
DualDVDRB - Dual core support for DVD-RB free
Rumbah is offline   Reply With Quote
Old 1st March 2019, 05:40   #1500  |  Link
Zebulon84
Registered User
 
Join Date: Apr 2015
Posts: 13
In this document (Final Call for Proposals for a Next-Generation Image Coding Standard (JPEG XL)) issued in last April, it is stated page 3:
Quote:
This new JPEG activity aims to develop a new image coding standard that provides state-of-the-art image compression performance, and that addresses shortcomings in current standards. To encourage widespread adoption, an important goal for this standard is to support a royalty-free baseline
So probably partially. But will il be enough to have wider adoption than previous standards (JPEG 2000, JPEG XR...) ?
Zebulon84 is offline   Reply With Quote
Reply

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 19:50.


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