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 > MPEG-4 AVC / H.264

Reply
 
Thread Tools Search this Thread Display Modes
Old 29th March 2013, 20:16   #21  |  Link
Dark Shikari
x264 developer
 
Dark Shikari's Avatar
 
Join Date: Sep 2005
Posts: 8,689
I doubt most of such claims, as they're often not substantiated and people who claim them usually refuse to post any actual tests.

However, it is important to consider that x264 is mainly optimized for at-least-mildly-lossy encoding. Broadcast television, encoding a web video, cloud gaming, videoconferencing, backing up a Blu-ray: all these are typical applications where you care somewhat about bitrate and will accept at least some small modicum of loss.

However, when encoding a Blu-ray, you often have the case where you have an insane amount of bitrate available, and all you care about is every pixel of every frame being as close to flawless as possible. Often in this case you don't actually need very good compression to get good results; it's more about consistency than anything else. And satisfying someone going over the video frame-by-frame is harder than someone looking at it in motion, since x264 judges quality based on temporal characteristics.

An interesting project for x264 would be to design a ratecontrol/RD setup based around just-noticeable-differences, to try to minimize visible artifacts at very high bitrates, instead of a goal of overall quality.

I still suspect x264 is the best option (quality-wise) for a lot of Blu-ray encoding, but I can certainly believe that another encoder, designed around eliminating any noticeable artifact at very high bitrates, could do a better job in some of those cases -- and with 30mbps on a relatively easy film, you hardly need the power of x264 to get good results.

Also, some commercial encoders basically oblige you to spend hours marking out areas that need higher quality during the encoding process, by going over the encoded video painstakingly -- naturally, the human eye can make vastly better decisions than x264 in that regard! You could make such a GUI with x264, too, but I don't think anyone has, so if you want to do something like that, you'll have to use the existing software.

Last edited by Dark Shikari; 29th March 2013 at 20:20.
Dark Shikari is offline   Reply With Quote
Old 29th March 2013, 20:31   #22  |  Link
jq963152
Registered User
 
Join Date: Apr 2012
Posts: 691
Quote:
Originally Posted by Dark Shikari View Post
An interesting project for x264 would be to design a ratecontrol/RD setup based around just-noticeable-differences, to try to minimize visible artifacts at very high bitrates, instead of a goal of overall quality.
Are you going to do this?
jq963152 is offline   Reply With Quote
Old 29th March 2013, 20:57   #23  |  Link
mandarinka
Registered User
 
mandarinka's Avatar
 
Join Date: Jan 2007
Posts: 738
From the wording, you can clearly see that it is not on the todo list
mandarinka is offline   Reply With Quote
Old 29th March 2013, 21:05   #24  |  Link
jq963152
Registered User
 
Join Date: Apr 2012
Posts: 691
Well, but it could also be interpreted as him thinking about putting it on the todo list .
jq963152 is offline   Reply With Quote
Old 29th March 2013, 21:25   #25  |  Link
Dark Shikari
x264 developer
 
Dark Shikari's Avatar
 
Join Date: Sep 2005
Posts: 8,689
It's a relatively hard problem, and might be hard to integrate into x264; it might or might not happen -- and even if someone does it, it might or might not work very well.
Dark Shikari is offline   Reply With Quote
Old 30th March 2013, 16:58   #26  |  Link
jq963152
Registered User
 
Join Date: Apr 2012
Posts: 691
Quote:
Originally Posted by Dark Shikari View Post
It's a relatively hard problem, and might be hard to integrate into x264; it might or might not happen -- and even if someone does it, it might or might not work very well.
Hmm, that's a bit unfortunate.

Because i am a little disappointed about this (especially the bolded parts):


Quote:
Originally Posted by Dark Shikari View Post
However, it is important to consider that x264 is mainly optimized for at-least-mildly-lossy encoding. Broadcast television, encoding a web video, cloud gaming, videoconferencing, backing up a Blu-ray: all these are typical applications where you care somewhat about bitrate and will accept at least some small modicum of loss.
Quote:
Originally Posted by Dark Shikari View Post
I still suspect x264 is the best option (quality-wise) for a lot of Blu-ray encoding, but I can certainly believe that another encoder, designed around eliminating any noticeable artifact at very high bitrates, could do a better job in some of those cases -- and with 30mbps on a relatively easy film, you hardly need the power of x264 to get good results.



Are there any other free H.264 encoders available which would be better for that than x264?

And, by the way, is there any thorough list available which lists all existing H.264 encoders?
jq963152 is offline   Reply With Quote
Old 30th March 2013, 16:59   #27  |  Link
Sagittaire
Testeur de codecs
 
Sagittaire's Avatar
 
Join Date: May 2003
Location: France
Posts: 2,420
Quote:
Originally Posted by Dark Shikari View Post

However, it is important to consider that x264 is mainly optimized for at-least-mildly-lossy encoding. Broadcast television, encoding a web video, cloud gaming, videoconferencing, backing up a Blu-ray: all these are typical applications where you care somewhat about bitrate and will accept at least some small modicum of loss.

However, when encoding a Blu-ray, you often have the case where you have an insane amount of bitrate available, and all you care about is every pixel of every frame being as close to flawless as possible. Often in this case you don't actually need very good compression to get good results; it's more about consistency than anything else. And satisfying someone going over the video frame-by-frame is harder than someone looking at it in motion, since x264 judges quality based on temporal characteristics.

An interesting project for x264 would be to design a ratecontrol/RD setup based around just-noticeable-differences, to try to minimize visible artifacts at very high bitrates, instead of a goal of overall quality.

I still suspect x264 is the best option (quality-wise) for a lot of Blu-ray encoding, but I can certainly believe that another encoder, designed around eliminating any noticeable artifact at very high bitrates, could do a better job in some of those cases -- and with 30mbps on a relatively easy film, you hardly need the power of x264 to get good results.

Also, some commercial encoders basically oblige you to spend hours marking out areas that need higher quality during the encoding process, by going over the encoded video painstakingly -- naturally, the human eye can make vastly better decisions than x264 in that regard! You could make such a GUI with x264, too, but I don't think anyone has, so if you want to do something like that, you'll have to use the existing software.
Yes ... exactly.

IMO x264 has really good design for" low/medium bitrate". x264 is IMO the best for overall quality result simply because x264 is optimized and designed for this.

For BD encoding you have other important parameters like "constant visual quality". For BD, overall bitrate is not problem. For exemple you can encode slow motion at 20-30 Mbps and for these part use bframe (for exemple) is not really good idea (bframe produce best overall quality but not best visual constant quality). Anyway you could have hard VBV constraint in high motion part (and/or perhaps hard noise constraint). In these particular part x264 will make really good job (and offen best job) because "best overall quality" will became more important than "best constant quality". If you want make BD encoding with x264, you must use really particular profil (low ratio for Iframe and Bframe for exemple).
__________________
Le Sagittaire ... ;-)

1- Ateme AVC or x264
2- VP7 or RV10 only for anime
3- XviD, DivX or WMV9
Sagittaire is offline   Reply With Quote
Old 30th March 2013, 17:08   #28  |  Link
jq963152
Registered User
 
Join Date: Apr 2012
Posts: 691
Quote:
Originally Posted by Sagittaire View Post
For BD encoding you have other important parameters like "constant visual quality".
What? Other important parameters? IMHO "constant visual quality" is the most important parameter (constant best visual quality to be precise ). Not just for Blu-ray Disc encoding, but for every encodng.

I'm a bit baffled/disappointed now to learn that x264 actually is not designed/optimized for this?

Quote:
Originally Posted by Sagittaire View Post
In these particular part x264 will make really good job (and offen best job) because "best overall quality" will became more important than "best constant quality".
IMHO:

Best overall quality = Best constant quality

Last edited by jq963152; 30th March 2013 at 17:14.
jq963152 is offline   Reply With Quote
Old 30th March 2013, 22:01   #29  |  Link
Dark Shikari
x264 developer
 
Dark Shikari's Avatar
 
Join Date: Sep 2005
Posts: 8,689
x264 is designed for constant visual quality, and IMO it's light-years better at this than any other encoder (at normal bitrates). At normal bitrates, for example, you want to raise the quantizer in high-motion areas because they are less perceptible to the viewer than an unmoving area -- "constant" visual quality is actually worse (static) quality in one than the other, because artifacts are more visible in one area than the other.

However, "constant visual quality" works very differently at extremely high rates. Doing the same thing as before might not have the same effect, because it could turn a scene where "everything is visually perfect" into a scene where "everything is visually perfect except for some of the high motion part". The extra bits given to the low motion part did make it higher quality, but not in a way that was very visible. This is exacerbated by encoders trying to make every frame individually perfect, since x264's temporal algorithms are counterproductive when the person judging quality is looking at single frames.
Dark Shikari is offline   Reply With Quote
Old 30th March 2013, 22:38   #30  |  Link
ChiDragon
Registered User
 
ChiDragon's Avatar
 
Join Date: Sep 2005
Location: Vancouver
Posts: 609
I've been reading Blu-ray Disc Demystified, and it brings up the point that the introduction of lossless audio formats turned the encoding workflow on its head. With DVD you always have CBR audio tracks, so you just subtract that combined bitrate from the bandwidth available to your video track and it doesn't matter what order you encode the video and audio tracks in. Now you encode the audio tracks first (and some discs have multiple lossless audio tracks produced by different encoders), then constrain your video encoding based on what the audio encoders have forced you to do. When you have big audio bitrate peaks, there's nothing you can do to make them smaller since these are lossless tracks we are talking about. You just have to lower your video bitrate at that point and try to maintain the quality.

This is all based on my reading into the book, so take it with a grain of salt, but it seems that dropping the current x264 into an authoring suite geared towards maximizing the video quality near the 48Mbps ceiling wouldn't work very well.
ChiDragon is offline   Reply With Quote
Old 30th March 2013, 23:36   #31  |  Link
Conquerist
Registered User
 
Join Date: Apr 2007
Posts: 37
Quote:
Originally Posted by Dark Shikari View Post
This is exacerbated by encoders trying to make every frame individually perfect, since x264's temporal algorithms are counterproductive when the person judging quality is looking at single frames.
I'm getting a little of topic here, but there was a recent paper (one of the authors is the guy behind SSIM) which basically said: HEVC looks better than AVC when played back as video, however every single frame in the clip, looked at individually, looks better in AVC. So, if you judge the video by looking at frames individually, HEVC is "worse" than AVC...

Of course the results of the comparison depend on the encoder used and not only on the standard, but in that comparison both encoders optimized for the same metric (PSNR), so it can be considered "fair" by some measure.

Last edited by Conquerist; 30th March 2013 at 23:38.
Conquerist is offline   Reply With Quote
Old 31st March 2013, 00:22   #32  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 3,089
Quote:
Originally Posted by ChiDragon View Post
I've been reading Blu-ray Disc Demystified, and it brings up the point that the introduction of lossless audio formats turned the encoding workflow on its head. With DVD you always have CBR audio tracks, so you just subtract that combined bitrate from the bandwidth available to your video track and it doesn't matter what order you encode the video and audio tracks in. Now you encode the audio tracks first (and some discs have multiple lossless audio tracks produced by different encoders), then constrain your video encoding based on what the audio encoders have forced you to do. When you have big audio bitrate peaks, there's nothing you can do to make them smaller since these are lossless tracks we are talking about. You just have to lower your video bitrate at that point and try to maintain the quality.
Well, BD has a maximum total mux rate of 54 Mbps IIRC, but a maximum video bitrate of 40 Mbps. So you'd need at least 14 Mbps of peak total audio + other stuff before you'd even have to modify peak bitrate. There are certainly discs like that, but it's not a common scenario.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 31st March 2013, 01:04   #33  |  Link
mp3dom
Registered User
 
Join Date: Jul 2003
Location: Italy
Posts: 1,044
The maximum data rate per transport stream is 48 Mbps, with 40 Mbps as a maximum data rate for the video (so there are 8 Mbps left for the audio without affecting the video bitrate). It is suggested to set a lower maximum bitrate for video (38 Mbps for example) to avoid possible buffer problems during multiplexing.
mp3dom is offline   Reply With Quote
Old 31st March 2013, 08:31   #34  |  Link
ChiDragon
Registered User
 
ChiDragon's Avatar
 
Join Date: Sep 2005
Location: Vancouver
Posts: 609
The Eurasian Prometheus has 5248kbps of CBR audio alone, plus the 4984kbps VBR main audio and 902kbps of subtitles (as you know, the equivalent of two DVD-quality surround tracks). For the big studios pushing out new releases, it's more common than you might think.

Perhaps the next big innovation will be to compress all the dub variants into one package since most of the channels are going to be duplicates most of the time. Even if packaged media is dead, it could reduce storage costs on the server side for dozens of languages while only one selection is sent over the wire.
ChiDragon is offline   Reply With Quote
Old 31st March 2013, 13:32   #35  |  Link
kieranrk
Registered User
 
Join Date: Jun 2009
Location: London, United Kingdom
Posts: 713
To answer the original question Ateme claim their latest software encoder beats x264:
http://www.olvitech.com/images/pdfs/...Technology.pdf

People who have done visual comparisons say both encoders are similar visually.
kieranrk is offline   Reply With Quote
Old 31st March 2013, 13:44   #36  |  Link
jq963152
Registered User
 
Join Date: Apr 2012
Posts: 691
Quote:
Originally Posted by Dark Shikari View Post
x264 is designed for constant visual quality, and IMO it's light-years better at this than any other encoder (at normal bitrates). At normal bitrates, for example, you want to raise the quantizer in high-motion areas because they are less perceptible to the viewer than an unmoving area -- "constant" visual quality is actually worse (static) quality in one than the other, because artifacts are more visible in one area than the other.

However, "constant visual quality" works very differently at extremely high rates. Doing the same thing as before might not have the same effect, because it could turn a scene where "everything is visually perfect" into a scene where "everything is visually perfect except for some of the high motion part". The extra bits given to the low motion part did make it higher quality, but not in a way that was very visible. This is exacerbated by encoders trying to make every frame individually perfect, since x264's temporal algorithms are counterproductive when the person judging quality is looking at single frames.

So, with other words: x264 is not designed/not optimized for constant visual quality at high bitrates?
jq963152 is offline   Reply With Quote
Old 31st March 2013, 20:48   #37  |  Link
Dark Shikari
x264 developer
 
Dark Shikari's Avatar
 
Join Date: Sep 2005
Posts: 8,689
Quote:
Originally Posted by jq963152 View Post
So, with other words: x264 is not designed/not optimized for constant visual quality at high bitrates?
That's a little bit of a strong statement, as lots of effort has gone into optimization at high rates; it's just that it's not the place where x264 has the greatest advantage.
Dark Shikari is offline   Reply With Quote
Old 31st March 2013, 21:35   #38  |  Link
Blue_MiSfit
Derek Prestegard IRL
 
Blue_MiSfit's Avatar
 
Join Date: Nov 2003
Location: Los Angeles
Posts: 5,620
I'll be taking a look at Ateme again this year
Blue_MiSfit is offline   Reply With Quote
Old 1st April 2013, 03:27   #39  |  Link
jq963152
Registered User
 
Join Date: Apr 2012
Posts: 691
Quote:
Originally Posted by Dark Shikari View Post
That's a little bit of a strong statement, as lots of effort has gone into optimization at high rates; it's just that it's not the place where x264 has the greatest advantage.
Well, this for example:

Quote:
Originally Posted by Dark Shikari View Post
since x264's temporal algorithms are counterproductive when the person judging quality is looking at single frames.

sounded more like a disadvantage rather than "not the greatest advantage".
jq963152 is offline   Reply With Quote
Old 1st April 2013, 05:32   #40  |  Link
Stephen R. Savage
Registered User
 
Stephen R. Savage's Avatar
 
Join Date: Nov 2009
Posts: 344
Quote:
Originally Posted by jq963152 View Post
Well, this for example:




sounded more like a disadvantage rather than "not the greatest advantage".
You can't judge image quality by looking at individual frames.
Stephen R. Savage 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 07:27.


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