"Codec Wars" : The attempt of an objective AVC/h.264 versus VC-1 benchmark - Page 13 - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #361 of 891 Old 04-05-2008, 01:44 AM
AVS Special Member
 
Joe Bloggs's Avatar
 
Join Date: Apr 2007
Location: UK
Posts: 2,534
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 111 Post(s)
Liked: 55
/\\ while grain can by simulated on cartoons, and it would be good for some tests, I think actual real film & HD video content (ie. non-cartoon/CGI) should be used as well. This is because, although CGI can be more controlled, it won't look/be the same as filming stuff with real cameras. One codec could be better at encoding CGI content (even with added noise/grain) but the other one could be better/more accurate at "real" camera recorded content.
Joe Bloggs is offline  
Sponsored Links
Advertisement
 
post #362 of 891 Old 04-05-2008, 02:04 AM
Member
 
Golgot13's Avatar
 
Join Date: Jun 2006
Location: Europe
Posts: 138
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Yes, I'm agree about CGI video but we need a full HD movie.
Ben can you share the FHD version of your video.
We will see the encoded file on BD or HDDVD player box.
Golgot13 is offline  
post #363 of 891 Old 04-07-2008, 05:32 AM
Advanced Member
 
tteich's Avatar
 
Join Date: Jan 2007
Posts: 784
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Golgot13 View Post

Yes, I'm agree about CGI video but we need a full HD movie.
Ben can you share the FHD version of your video.
We will see the encoded file on BD or HDDVD player box.

If I only had an analog HD video capture card in my hands, then I'd capture a scene from one of my countless HiVision LDs. I'd guess there's enough random noise in them to be a good analog source for encoding tests.

Can anyone recommend a (not too pricey) HD YUV capturing solution (PCI card, or external box)?

Laserdiscs: ~350
HiVision LDs: 42
HD DVD count: 379 / still 20 on the wishlist
tteich is offline  
post #364 of 891 Old 04-07-2008, 06:17 AM
Member
 
Golgot13's Avatar
 
Join Date: Jun 2006
Location: Europe
Posts: 138
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Hi Ben,

Quote:
Originally Posted by benwaggoner View Post

Yes, I just posted it to show what the content looks like, so we can get some validation that it's a reasonable test.

I wait full HD format version of your video...


Quote:
Originally Posted by benwaggoner View Post

I don't want to do all the work on the clip and have people argue it's somehow not a valid test after the fact. I want to get buy in about what needs to be in there to make it a valid test.

No problem, we will use "for you" the last core of PeP/CInevisionPSE (1.07a or 1.07b).
We will post all data (options, preset, metrics: PSNR, vbv,...) not like your test on your old website page about VC1 and H264...

I really want to make the comparison before NAB but I think you don't want...


Today I'm sure that x264 can produce better result than most HD studio encoder solution for BD (or HDDVD).
Golgot13 is offline  
post #365 of 891 Old 04-07-2008, 06:53 AM
Member
 
Golgot13's Avatar
 
Join Date: Jun 2006
Location: Europe
Posts: 138
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by tteich View Post

Can anyone recommend a (not too pricey) HD YUV capturing solution (PCI card, or external box)?

BlackMagic card, it use by professional but the price is low (start at 1000$):

http://www.blackmagic-design.com/products/hd/


Your video will be nice to test interlaced source but not progressive like video from Ben (filmsource).
Golgot13 is offline  
post #366 of 891 Old 04-07-2008, 07:48 AM
Advanced Member
 
tteich's Avatar
 
Join Date: Jan 2007
Posts: 784
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Golgot13 View Post

BlackMagic card, it use by professional but the price is low (start at 1000$):

http://www.blackmagic-design.com/products/hd/


Your video will be nice to test interlaced source but not progressive like video from Ben (filmsource).

Thanks Golgot13. Will take a look at this card.

By the way: do you have the x264 encoder settings handy, which allows to produce a HDDVD compliant elementary stream?

Laserdiscs: ~350
HiVision LDs: 42
HD DVD count: 379 / still 20 on the wishlist
tteich is offline  
post #367 of 891 Old 04-07-2008, 09:21 AM
Member
 
Golgot13's Avatar
 
Join Date: Jun 2006
Location: Europe
Posts: 138
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by tteich View Post

Thanks Golgot13. Will take a look at this card.
By the way: do you have the x264 encoder settings handy, which allows to produce a HDDVD compliant elementary stream?

See the setting in thread "Codec Challenge" (for HDDVD and BD) on Doom9:
http://forum.doom9.org/showthread.php?t=114928

But you need a special version of x264 (for vbv compliant) and you will need to patch the stream with H264info
(Trahald from Doom9 make a good work, I showed him for PPS and 3:2 pulldown flag to convert H264 stream: HDDVD <-> BD ).
Golgot13 is offline  
post #368 of 891 Old 04-07-2008, 11:46 AM
Advanced Member
 
tteich's Avatar
 
Join Date: Jan 2007
Posts: 784
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Golgot13 View Post

See the setting in thread "Codec Challenge" (for HDDVD and BD) on Doom9:
http://forum.doom9.org/showthread.php?t=114928

But you need a special version of x264 (for vbv compliant) and you will need to patch the stream with H264info
(Trahald from Doom9 make a good work, I showed him for PPS and 3:2 pulldown flag to convert H264 stream: HDDVD <-> BD ).

Thanks for pointing me to the right thread, that's exactly what I'm looking for!

Laserdiscs: ~350
HiVision LDs: 42
HD DVD count: 379 / still 20 on the wishlist
tteich is offline  
post #369 of 891 Old 04-07-2008, 11:51 AM
AVS Special Member
 
benwaggoner's Avatar
 
Join Date: Sep 2003
Location: Portland, OR
Posts: 2,616
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Golgot13 View Post

...But you need a special version of x264 (for vbv compliant) and you will need to patch the stream with H264info
(Trahald from Doom9 make a good work, I showed him for PPS and 3:2 pulldown flag to convert H264 stream: HDDVD <-> BD ).

Which is why no one is using it for actual professional disc production.

There's a big difference between a technology and a product. We have to do a lot of beta testing and validation of a particular build before we certify it as being compatible with particular players and specs.

Digital Media Technology Insider with Microsoft

My compression blog
benwaggoner is offline  
post #370 of 891 Old 04-08-2008, 01:17 AM
Member
 
Golgot13's Avatar
 
Join Date: Jun 2006
Location: Europe
Posts: 138
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by benwaggoner View Post

Which is why no one is using it for actual professional disc production.

This version is not public, it's a special version compiled by specialist of x264.
I used it on some HDDVD project (more than 2), it work well on Toshiba players.
And same video stream work well on BD players (after patch NAL/header of H264 stream).

Quote:
Originally Posted by benwaggoner View Post

There's a big difference between a technology and a product. We have to do a lot of beta testing and validation of a particular build before we certify it as being compatible with particular players and specs.


About your quality of your test, I remember that one guy from Doom9 show you
there was some "vbv" problem (and quality of it) with your PeP 1.06 (MS release a new version 1.07...).

On via and doom9 communities, they do too many test: you will surprise
if you know how many professionel test their plateform with x264 stream...
Today, x264 can generate a fully compliant stream with BD or HDDVD ( and DVB, IPTV, 3G, ISDB, DMB,...).


Put your video, Ben, and we will show you !
Golgot13 is offline  
post #371 of 891 Old 04-08-2008, 04:41 AM
AVS Addicted Member
 
trbarry's Avatar
 
Join Date: Jan 2000
Location: Gainesville FL USA
Posts: 10,138
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Golgot13 View Post

...
Put your video, Ben, and we will show you !

That does pretty well sum it up. I would like to see this.

- Tom

Why don't we power our electric cars from greener, cheaper Liquid Fluoride Thorium Reactors?

Tom Barry - Find my video filters at www.trbarry.com
trbarry is offline  
post #372 of 891 Old 04-09-2008, 02:59 AM
Member
 
Golgot13's Avatar
 
Join Date: Jun 2006
Location: Europe
Posts: 138
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by trbarry View Post

That does pretty well sum it up. I would like to see this.

- Tom


Same, but it will not be possible this week:
Ben participate at DVDA seminar before NAB this saturday. But I don't understand why he will talk about H264...
"Encoding for the Next Generation Discs:: MPEG-2/H.264/VC1
Presenter: Ben Waggoner, Senior Program Manager, Video Encoding, Microsoft"

I understand why, in US market, authoring studio prefer VC1. I'm sure people on this seminar will not see and know
the truth about H264 or MPEG2 on next generation disc...


I will really happy if some people of DVDA could see the video from Ben encoded in H264 before the seminar.
And I will be happy if some people can talk about this thread and the Doom9's thread (Codec's challenge) with Ben on DVDA seminar...

Last, I think today MS will not participate at next MSU (Moscow State University) video codec comparison to show the VC1 potential.
Last year they said me I talked late about it (I said it in May, MSU close the participation in August)...
Golgot13 is offline  
post #373 of 891 Old 04-09-2008, 07:17 AM
AVS Special Member
 
PaulGo's Avatar
 
Join Date: Apr 2003
Location: North Potomac, MD
Posts: 3,626
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 8 Post(s)
Liked: 19
Thank you for doing a very effective job of informing people and getting out the truth!

Quote:
Originally Posted by Golgot13 View Post

Same, but it will not be possible this week:
Ben participate at DVDA seminar before NAB this saturday. But I don't understand why he will talk about H264...
"Encoding for the Next Generation Discs:: MPEG-2/H.264/VC1
Presenter: Ben Waggoner, Senior Program Manager, Video Encoding, Microsoft"

I understand why, in US market, authoring studio prefer VC1. I'm sure people on this seminar will not see and know
the truth about H264 or MPEG2 on next generation disc...


I will really happy if some people of DVDA could see the video from Ben encoded in H264 before the seminar.
And I will be happy if some people can talk about this thread and the Doom9's thread (Codec's challenge) with Ben on DVDA seminar...

Last, I think today MS will not participate at next MSU (Moscow State University) video codec comparison to show the VC1 potential.
Last year they said me I talked late about it (I said it in May, MSU close the participation in August)...

PaulGo is offline  
post #374 of 891 Old 04-10-2008, 12:40 PM
AVS Special Member
 
benwaggoner's Avatar
 
Join Date: Sep 2003
Location: Portland, OR
Posts: 2,616
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by trbarry View Post

That does pretty well sum it up. I would like to see this.

- Tom

Well, what I was expecting to happen is that I'd post a rough cut and people would give me some feedback on whether or not it was appropriate content for a test, and if not, what sorts of modifications would make it a better test. As it is, I've gotten about a dozen posts complaining it isn't done for every post actually trying to help the process along.

What I don't want to happen is to spend the 40-50 hours it'll take to really finish it up right and then have people tell me after the fact why it's not an appropriate test. I really want to do this as a useful community resource, not something to get turned into AVS gotcha sniping.

Yes, it's taken me a long time to get it done (and note that I'm not even on the codec team anymore), but I want to finish it up. So, for those of you anxious to see it, please, give me feedback about the rough cut on the Doom9 thread!

http://forum.doom9.org/showthread.php?t=135938

Digital Media Technology Insider with Microsoft

My compression blog
benwaggoner is offline  
post #375 of 891 Old 04-21-2008, 12:26 PM
AVS Special Member
 
benwaggoner's Avatar
 
Join Date: Sep 2003
Location: Portland, OR
Posts: 2,616
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Golgot13 View Post

NAB is finish, so we can restart the thread to finish the challenge.
I wait the HD video....

Sure. But do you want to weigh in on modifications to the test clip? I don't think I've see any detailed response from you on that.

Quote:


I surprise by all news from you at NAB and DVDA.
You said on this thread we can not compare VC1 and H264,
because many people (me first or second) say H264 is better than VC1.
But my feedback from DVDA/NAB is: you said VC1 give better result than H264
at same bitrate !!!!!
Can you show your videos and explain what software and preset did you use?

I think what you heard reported is my quoting a number of Hollywood compressionists and directors who have been comparing the latest PEP and competing H.264 encodres, who said that VC-1 is doing a much better job of preserving film grain texture and avoding blocking in blacks.

This isn't just from my testing, but a variety of studios doing A/B compares with their own in-progress Blu-ray titles. These conversations were under NDA, of course, but I'll see if I can get some attributable quotes.

Digital Media Technology Insider with Microsoft

My compression blog
benwaggoner is offline  
post #376 of 891 Old 04-22-2008, 05:30 AM
Member
 
Golgot13's Avatar
 
Join Date: Jun 2006
Location: Europe
Posts: 138
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by benwaggoner View Post

Sure. But do you want to weigh in on modifications to the test clip? I don't think I've see any detailed response from you on that.

I posted a answer:
http://forum.doom9.org/showthread.ph...39#post1124639

We test the codec efficiency so put what you want because we don't judge the video quality
(colour, noise, movie grain, effect, ...).
Yes the video quality will influence the result but I'm sure of H264 codec.


Quote:
Originally Posted by benwaggoner View Post

I think what you heard reported is my quoting a number of Hollywood compressionists and directors who have been comparing the latest PEP and competing H.264 encodres, who said that VC-1 is doing a much better job of preserving film grain texture and avoding blocking in blacks.

Yes, but I'm not sure that you use the last encoder or the best preset for H264 encoding process...
I'm sure for VC1 encoding process you take much time than H264 encoding (not same pre-processing,...)
You can put the name of H264 encoder and his preset that you use (you don't like sometime to answer a specific ask ...)

Quote:
Originally Posted by benwaggoner View Post

This isn't just from my testing, but a variety of studios doing A/B compares with their own in-progress Blu-ray titles. These conversations were under NDA, of course, but I'll see if I can get some attributable quotes.

I don't listen this...
You show the difference between your VC1 encoding (with pre-process) and a H264 encoding.
Strange you don't invite any company which make a H264 encoder like Mainconcept (Sonic use their SDK),
CTC, Thomson, Sony (two japaneses present at NAB and not the guy who make a demo at DVDA), ...

VC1 will die soon because H264 HD hardware encoder appear which encode in real time (I find one which can make a good quality,
because 20Mbps is enough...) and no company developp a good VC1 (except Inlet which is not compliant with BD or HDDVD).
Golgot13 is offline  
post #377 of 891 Old 04-22-2008, 05:49 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,022
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 695 Post(s)
Liked: 397
Quote:
Originally Posted by Golgot13 View Post

VC1 will die soon because H264 HD hardware encoder appear which encode in real time (I find one which can make a good quality,
because 20Mbps is enough...) and no company developp a good VC1 (except Inlet which is not compliant with BD or HDDVD).

It woud be a sad, sad day if real-time encoders are used for Blu-ray. That would put speed over quality. Such devices are very inflexible and difficult to improve. With speed on their side, people will just use them as is and we would be locked in time. Such was the case with MPEG-2 encoders....

Amir
Founder, Madrona Digital
"Insist on Quality Engineering"

amirm is offline  
post #378 of 891 Old 04-22-2008, 07:16 AM
Member
 
Golgot13's Avatar
 
Join Date: Jun 2006
Location: Europe
Posts: 138
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by amirm View Post

It woud be a sad, sad day if real-time encoders are used for Blu-ray. That would put speed over quality. Such devices are very inflexible and difficult to improve. With speed on their side, people will just use them as is and we would be locked in time. Such was the case with MPEG-2 encoders....

???
On last version PeP, there is more optimization for speed encoding than quality encoding (support more core, ...).
And you correct some bug, some bug find by Doom9 guy (vbv...) to increase the PeP VC1 quality.


Amirm, may be you can answer at my ask: what H264 encoder Ben use for DVDA demo and why no H264 company is invited
to discuss about video quality (it's easy to show what you want if there is nobody to say and show it is not true).


I hope to see one day your HD source to show.
It's very simple to finish and know who say true: put the video source...
Golgot13 is offline  
post #379 of 891 Old 04-22-2008, 08:04 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,022
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 695 Post(s)
Liked: 397
Quote:
Originally Posted by Golgot13 View Post

On last version PeP, there is more optimization for speed encoding than quality encoding (support more core, ...).

Making things faster is good. Hardwiring them in some real-time box, using specialized DSPs and hard to program FPGAs, not.

Quote:


And you correct some bug, some bug find by Doom9 guy (vbv...) to increase the PeP VC1 quality.

I am sure Microsoft is in the process of sending him a free cookie for finding a bug .


Quote:


Amirm, may be you can answer at my ask: what H264 encoder Ben use for DVDA demo and why no H264 company is invited
to discuss about video quality (it's easy to show what you want if there is nobody to say and show it is not true).

I have no idea. I left Microsoft in January and don't keep up with stuff like this.

Quote:


I hope to see one day your HD source to show.
It's very simple to finish and know who say true: put the video source...

The message that HD optical encoding invovles manual intervention always seems lost in arguments in this thread. You don't use a compact car to race around the track, nor use the latter to go to work everyday. Just because ever car has 4 wheels, doesn't mean they are all the same. Automated encoding using random tools here and there is of no interest or value to people creating titles people care about here. For personal archiving, sure. But not for HD optical.

Amir
Founder, Madrona Digital
"Insist on Quality Engineering"

amirm is offline  
post #380 of 891 Old 04-22-2008, 09:57 AM
Member
 
Golgot13's Avatar
 
Join Date: Jun 2006
Location: Europe
Posts: 138
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by amirm View Post

I am sure Microsoft is in the process of sending him a free cookie for finding a bug .

No, because the guy has not officially the soft...
But it's only to prouve MS make sometime mistakes...


Quote:
Originally Posted by amirm View Post

The message that HD optical encoding invovles manual intervention always seems lost in arguments in this thread. You don't use a compact car to race around the track, nor use the latter to go to work everyday. Just because ever car has 4 wheels, doesn't mean they are all the same. Automated encoding using random tools here and there is of no interest or value to people creating titles people care about here. For personal archiving, sure. But not for HD optical.

On the thread we talk about codec with HD optical restriction, so we can use any tool to my mind.
But you're right I can discuss about H264 studio professional encoder from some companies (I know 4 different package)
Some of this encoder can do better quality than VC1 from PeP, and have many pre-processing and preset for BD use....
Golgot13 is offline  
post #381 of 891 Old 04-22-2008, 10:14 AM
AVS Special Member
 
Neo1965's Avatar
 
Join Date: Jun 2006
Location: CANADA!
Posts: 3,525
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Quote:
Originally Posted by amirm View Post

Automated encoding using random tools here and there is of no interest or value to people creating titles people care about here. For personal archiving, sure. But not for HD optical.

You're right on DSPs not being the way to go, DSPs are too limited in their capabilities. FPGAs are useful only for verification as part of the answer, they are too slow to run real time.

Once you have a strong enough algorithm, you can go to verilog and solve your problem there. That and a mill or two should about do it to beat any DSP or sw solution. As long as you find people willing to take these chips off your hands, even the mill or two can go away quickly.
---

I wouldn't call the tools used in automated encoding "random". They're likely field tested and often do have tricks not known to the outside world. Besides the wisdom of not sharing your best secrets to competitors, the people making these real time encoders used in broadcast don't often feel a need to get into a pissing contest with the riff-raff because the only opinions they need to worry about are how their customers who build broadcast equipment to process video think about the video quality.

FWIW, the x264 generated and even the hand tuned 'optimal' VC-1 encodes of Elephant Dreams are not quite excellent, and they're already breaking many of the limits on max bitrate imposed by the realities of broadcast to achieve their current quality, which can still be improved.
Neo1965 is offline  
post #382 of 891 Old 04-22-2008, 10:36 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,022
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 695 Post(s)
Liked: 397
Quote:
Originally Posted by Neo1965 View Post

You're right on DSPs not being the way to go, DSPs are too limited in their capabilities. FPGAs are useful only for verification as part of the answer, they are too slow to run real time.

Compared to software, FPGAs do provide significant speed up. I think you are comparing them to ASICs in which case, that is true. But often, FPGAs are used to implement a fixed algorithm faster while the rest is powered by a general pupose CPU (the one in a PC or otherwise).

Quote:


Once you have a strong enough algorithm, you can go to verilog and solve your problem there.

Given the low number of units sold for encoder, it is not usually economical to pay the NRE for ASICs. So the solution usually stays in the domain of DSP+FPGA or CPU+FPGA or some other similar permuation.

Amir
Founder, Madrona Digital
"Insist on Quality Engineering"

amirm is offline  
post #383 of 891 Old 04-22-2008, 11:09 AM
AVS Special Member
 
sspears's Avatar
 
Join Date: Feb 1999
Location: Sammamish, WA, USA
Posts: 5,244
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 51
Quote:


On last version PeP, there is more optimization for speed encoding than quality encoding (support more core, ...).

This is not true. PEP 2.1 is slower overall. Don't confuse PEP with the SDK. PEP is superior and has features the SDK does not support in terms of quality.

PEP supports the same # of cores as it always has, 4. 1.8x real-time is about that fastest PEP has ever encoded HD content. This was very expensive HP HW with a SAN that was well over $100k all by itself. Add in the cost of the HP HW and its more than my hosue.

BTW - VC-1 adoption is actually picking up on BD. A big shootout was done recently between Cinemcraft AVC, Sony AVC and PEP. PEP was the winner. At least this is the information that was given to us by those that did the shootout.
sspears is offline  
post #384 of 891 Old 04-22-2008, 11:20 AM
AVS Special Member
 
benwaggoner's Avatar
 
Join Date: Sep 2003
Location: Portland, OR
Posts: 2,616
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Golgot13 View Post

I posted a answer:
http://forum.doom9.org/showthread.ph...39#post1124639

We test the codec efficiency so put what you want because we don't judge the video quality
(colour, noise, movie grain, effect, ...).
Yes the video quality will influence the result but I'm sure of H264 codec.

The reason why I'm interested in doing that project is to create a test clip that's relevant to the AVS-centric goal of being able to encode Hollywood movies. So I want to make a clip that matches the mix of different kinds of feature content in a realistic way. I'm looking for feedback on what aspects of features we need to capture; edits, transitions, textures, lighting, etcetera.

You seem really excited to use this clip to prove a point, but I'd appreciate your and everyone's help and feedback in making it something that's actually useful and relevant for this community.

Digital Media Technology Insider with Microsoft

My compression blog
benwaggoner is offline  
post #385 of 891 Old 04-22-2008, 11:27 AM
AVS Special Member
 
MovieSwede's Avatar
 
Join Date: Feb 2007
Location: Gothenburg
Posts: 6,762
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 11 Post(s)
Liked: 37
Ben

Would a strobing clip like the high shutter scenes in Saving Private Ryan be harder to encode since every frame is much more on its own. Or does the abscense of motionblur help?

Good movies are as rare as an on topic discussion.
MovieSwede is offline  
post #386 of 891 Old 04-22-2008, 11:30 AM
AVS Special Member
 
benwaggoner's Avatar
 
Join Date: Sep 2003
Location: Portland, OR
Posts: 2,616
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Golgot13 View Post

???
On last version PeP, there is more optimization for speed encoding than quality encoding (support more core, ...).
And you correct some bug, some bug find by Doom9 guy (vbv...) to increase the PeP VC1 quality.

You seem to talk a lot about PEP, but it doesn't sound like you've been using recent versions.

The "multi-core" enhancement in the recent versions is a workflow tweak to allow it to run multiple instances on the same hardware, so two workers can be running together on an 8-core blade. There's been other work around supporting newer Intel/AMD instructions. But that's all about making existing algorithms faster, not figuring out how to reduce encoding time at the expense of speed, which is what real-time encoding is all about.

Quote:


Amirm, may be you can answer at my ask: what H264 encoder Ben use for DVDA demo and why no H264 company is invited
to discuss about video quality (it's easy to show what you want if there is nobody to say and show it is not true).

You can ask me, I'm right here .

The venue only had 1024x768 projectors, so we didn't do a lot of quality demos. I did show a few SD VOD clips to demonstrate how different encoders result in different quality, comparing Apple's and Main Concept's H.264 to VC-1, and how quality improvements are limited by display in the end.

Sony talked some about their new encoder, although I had to run off to a meeting and didn't hear it all. I'm not sure who else all presented; the schedule was pretty dynamic.

Digital Media Technology Insider with Microsoft

My compression blog
benwaggoner is offline  
post #387 of 891 Old 04-22-2008, 11:31 AM
AVS Special Member
 
benwaggoner's Avatar
 
Join Date: Sep 2003
Location: Portland, OR
Posts: 2,616
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by MovieSwede View Post

Would a strobing clip like the high shutter scenes in Saving Private Ryan be harder to encode since every frame is much more on its own. Or does the abscense of motionblur help?

In general short-shutter content is harder to encode due to the lack of motion blur, and also because it's typically grainy. Fast moving sharp edges are quite a challenge. If there are repeat frames, those definitely help, but flash frames can be challenging depending on the encoder (ours does really well). Partial flash frames can be the worst, where part of the frame is flashing and the rest isn't.

Digital Media Technology Insider with Microsoft

My compression blog
benwaggoner is offline  
post #388 of 891 Old 04-22-2008, 05:22 PM
AVS Special Member
 
benwaggoner's Avatar
 
Join Date: Sep 2003
Location: Portland, OR
Posts: 2,616
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Speaking of my NAB presentation, the actual deck I used is now up here:

http://www.on10.net/blogs/benwagg/22040/

Digital Media Technology Insider with Microsoft

My compression blog
benwaggoner is offline  
post #389 of 891 Old 04-22-2008, 06:49 PM
Advanced Member
 
darkedgex's Avatar
 
Join Date: Apr 2006
Location: WA
Posts: 734
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 19
I haven't seen any of the footage you've made available, but wouldn't some material similar to what was available on that Nine Inch Nails BD/HD DVD be useful as a torture test for encoders? As I recall, you (Ben) had indicated this material was particularly difficult. It's not representative of typical motion picture material, but it might be interesting to include.

Fight mediocrity: Insist on BD50 discs for all movies longer than 100 minutes, optimized video encodes that fully utilize the available space, lossless audio track, and new masters for catalog titles!
darkedgex is offline  
post #390 of 891 Old 04-22-2008, 06:56 PM
AVS Special Member
 
benwaggoner's Avatar
 
Join Date: Sep 2003
Location: Portland, OR
Posts: 2,616
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by darkedgex View Post

I haven't seen any of the footage you've made available, but wouldn't some material similar to what was available on that Nine Inch Nails BD/HD DVD be useful as a torture test for encoders? As I recall, you (Ben) had indicated this material was particularly difficult. It's not representative of typical motion picture material, but it might be interesting to include.

I actually have some 1080i source I can throw into the mix, and it's even actually complete! Interesting stuff, if there's interest in it. I've mainly focused on the 24p stuff since that's the AVS bread-and-butter.

Here's a SD downconvert of it I did earlier (I could do a lot better):

http://silverlight.services.live.com...l1/iframe.html

Digital Media Technology Insider with Microsoft

My compression blog
benwaggoner is offline  
Reply HDTV Software Media Discussion

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

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
Trackbacks are Off
Pingbacks are Off
Refbacks are Off