[HD 1080p24 Challenge] MPEG2, VC-1 and H264 with real uncompressed source

[Golgot13]

you should know how to make public the results of your encodings without risking them being traced back to you.

Yes, but the video used (with movie grain) is copyrighted...
And some people will say it's not true measure because they can not reproduce it(if they don't have this video).
We wait the video from Ben.

And the contact of Sagitaire use VC1enc.exe, I can make for H264 (like MS with VC1) my encoding option.


About FRExt, it was adopted in July 2004.

[benwaggoner]

Real life non-CG material (everybody is expecting Ben to provide) should answer the question about grain preservation and such...
Just to update, I have the source, but both my workstations are down right now :(. I'll probably just give up and start doing a proxy edit on my laptop tomorrow, and then conform when my Barcelona is finally up to snuff.

So, we're still a couple weeks out from something final. If folks are interested, I might make a SD test encode of it available earlier to give folks a taste of the source, cuts, etcetera. The goal is to make it "realistically challenging" so I welcome suggestions for the kinds of edits and such people would like in this kind of test clip.

[Sagittaire]

Just to update, I have the source, but both my workstations are down right now :(. I'll probably just give up and start doing a proxy edit on my laptop tomorrow, and then conform when my Barcelona is finally up to snuff.


Possible to make lossless encoding with high efficient codec (like FFV1 for example). With directly good color matrix (YV12 bt709). With directly the good pre-process (dithering, dark noise filtering ... etc etc).

[benwaggoner]

Possible to make lossless encoding with high efficient codec (like FFV1 for example). With directly good color matrix (YV12 bt709). With directly the good pre-process (dithering, dark noise filtering ... etc etc).
Yes, I'd be providing a 4:2:0 master for the testing, so we don't have to argue about dithering modes and such.

I've been considering doing either a .zip of an IYUV .avi file, or using Lagarith in its YV12 mode. Either preference?

Probably should start a new thread to discuss this...

[Dark Shikari]

Yes, I'd be providing a 4:2:0 master for the testing, so we don't have to argue about dithering modes and such.

I've been considering doing either a .zip of an IYUV .avi file, or using Lagarith in its YV12 mode. Either preference?

Probably should start a new thread to discuss this...
FFV1 is probably better than both, and its supported by libavcodec.

ZIP is also a very very bad compressor and will take ages to decompress.

[Golgot13]

Agree it will be better if we have FFV1 .

[Sharktooth]

h.264 lossless...

[Golgot13]

h.264 lossless...

Yes, but the speed to encode and decode will not be same :D.


@Ben:
there is not VC1 lossless option ?
VC1 will be on codec comparison 2007?
(I talked about it in june with you and Zambelli).

[Sharktooth]

well, if you have a new videocard the decoding can be HW assisted so you will have faster re-encoding... ;)

[Golgot13]

well, if you have a new videocard the decoding can be HW assisted so you will have faster re-encoding... ;)

Yes I have a Nvidia 8500 but I 'm not sure it can decode H264 with Lossless option (?)

About Fourth Annual MSU H264 comparison, there is some japanese company which is contributor, KDDI Labs
(made the first H264, up to HP@L4.1, editing software at frame accure). I hope to see NTT which have software encoder
(1920x1080 video in H264 at 3Mbps, demo at NAB !!!).
VC1 can be compare like Divx on 3rd Annual MSU...

[Inventive Software]

Sure it can decode it....... it's just an enormous bitrate over "normal" content is all.

[Golgot13]

Sure it can decode it....... it's just an enormous bitrate over "normal" content is all.

There is no H264 bitrate limitation of NVidia acceleration.
Because all hardware chipset decoder (ofr H264) are limited by profile and bitrate.

[benwaggoner]

FFV1 is probably better than both, and its supported by libavcodec.

ZIP is also a very very bad compressor and will take ages to decompress.
Fair enough. But I'm hoping to have a pretty universally playable file, and I think an IYUV .avi file will play back pretty much everywhere, even in QuickTime on a Mac.

One of the benefits of being at Microsoft is my download bandwidth isn't capped :).

Of course, since the content is provided for compression testing, everyone is welcome to use it to test and distribute with lossless codecs as well.

[Dark Shikari]

Fair enough. But I'm hoping to have a pretty universally playable file, and I think an IYUV .avi file will play back pretty much everywhere, even in QuickTime on a Mac.

One of the benefits of being at Microsoft is my download bandwidth isn't capped :).

Of course, since the content is provided for compression testing, everyone is welcome to use it to test and distribute with lossless codecs as well.An IYUV would be absolutely humongous though, and zipping it would be really pointless because while it would lower its size, ZIP is not designed for video contexts and as such is less efficient than the least efficient lossless codecs in my experience.

You could upload both an FFV1 for the sane people, and a ZIP for the insane people, and see how many people are in their right mind... :D

[zambelli]

No, when you give a software for free at one company (or 2 only) and help to make a product, you lock the market.
I think you misunderstand the principles of unfair competition. Locking a market means locking out and preventing companies from joining and competing in a market in order to preserve one's domination. Considering that Microsoft wasn't making any money on the VC-1 encoder and that it was in no way preventing authoring companies from encoding for HD-DVD using H.264, MPEG-2 or competing VC-1 encoders - I really don't see how that could even remotely be considered unfair competition.

Just because somebody didn't give you something for free doesn't automatically give you the right to have had it.

And MS choosed only two partner only in Europe about this.
All other company which asked to have this software (without support because MS can't support much company)
received no answer (I'm not alone for this).
We've been over this before. It wasn't a political decision and Microsoft wasn't trying to control the HD-DVD authoring market. It was a simple matter of prioritizing where limited resources should be spent.

VC1 will be on codec comparison 2007?
(I talked about it in june with you and Zambelli).
No, I'm afraid not. MSU extended us a last-minute invitation but we simply couldn't prepare in time for the comparison. It came right around the time we were releasing the VC-1 Encoder SDK and preparing for IBC.

[zambelli]

Fair enough. But I'm hoping to have a pretty universally playable file, and I think an IYUV .avi file will play back pretty much everywhere, even in QuickTime on a Mac.
We can make a lossless H.264 in MP4 video for the Mac folks. It'll be more useful than an AVI anyway.

[Golgot13]

Just because somebody didn't give you something for free doesn't automatically give you the right to have had it.


Like Windows Media Player on MS OS, your partner could have code, software and more.
But all other companies could do nothing, and lost part of market or all market.
EU commission was not agree with this (you know). And it was same for the market (HDDVD).
I think it's why there is a inquire of EU commission (and for other reasons)

I didn't know why we could not buy it too, I wasn't agree because it was impossible to have it only.

[Golgot13]

No, I'm afraid not. MSU extended us a last-minute invitation but we simply couldn't prepare in time for the comparison. It came right around the time we were releasing the VC-1 Encoder SDK and preparing for IBC.



But I think you can ask today to be included on this test because your codec have a big part
of HD disc market.

Do you read the comparison between picture codec test with MS HD Photo and different JPEG2000 codec?
http://www.compression.ru/video/codec_comparison/wmp_codecs_comparison_en.html

Or download the test in PDF format:
http://www.compression.ru/video/codec_comparison/pdf/wmp_codec_comparison_en.pdf

[Sharktooth]

http://forum.doom9.org/showthread.php?p=1047342#post1047342

[Sharktooth]

those things make me laugh...

[Golgot13]

Of course, since the content is provided for compression testing, everyone is welcome to use it to test and distribute with lossless codecs as well.

Ben, I wait the source (since more than one month).
About H264, DirecTV announce 21 more HD channels (in H264), so in USA (MS 's country)
people will can see 51 H264 HD channels and at the end of 2007 there will be 100 H264 HD channel.
For VC1 in IPTV, there is only Swisscom project (?), all others in SD use MPEG2 and H264.

[Sharktooth]

VC-1 is doomed. Let it die alone...

[benwaggoner]

VC-1 is doomed. Let it die alone...
In what context?

Given that VC-1 is preinstalled on 95% of the world's computers, it's rather hugely important. Were we to decide to spike it today (which we aren't :)), people would still be using it in huge volume years from now. Don't underestimate the slow pace of codec adoption outside in the general public - we're not even to the point where I consider using VC-1 Advanced Profile "safe" for a general audience unless Silverlight or MCE is being targeted, given how many WMP9 installs on managed desktops there are out there.

[Golgot13]

VC-1 is doomed. Let it die alone...

I'm agree with Ben about VC1 and computer, but it will be hard to be used by another "support".
H264 and MPEG2 are used on all others type. To my mind, VC1 is good for Handheld system
(need less CPU and consumption) but today some manufactor announce H264 chipset optimized
for Handheld devices (less consumption than chipset of today).


But Ben, it's possible to know when we could test our codecs on the source ? :rolleyes: :)

Can you post too your best VC1 encoding with this source ?

[benwaggoner]

I'm agree with Ben about VC1 and computer, but it will be hard to be used by another "support".
H264 and MPEG2 are used on all others type. To my mind, VC1 is good for Handheld system
(need less CPU and consumption) but today some manufactor announce H264 chipset optimized
for Handheld devices (less consumption than chipset of today).
You're forgetting that the majority of HD optical discs use VC-1!

But Ben, it's possible to know when we could test our codecs on the source ? :rolleyes: :)

Can you post too your best VC1 encoding with this source ?
I just got back the workstation yesterday afternoon (turns out it was a sound card conflict). Fingers crossed, I'll be able to start logging the video tomorrow.

[Golgot13]

You're forgetting that the majority of HD optical discs use VC-1!

Yes, sure for HDDVD but there is more and more exception with H264 (in UK, Germany, France,...)
and with BD, Sony prefer MPEG2 (bad implementation sometime) and H264, others H264 and some VC1.


I just got back the workstation yesterday afternoon (turns out it was a sound card conflict). Fingers crossed, I'll be able to start logging the video tomorrow.

I hope you will not use Vista (lot of problem with driver and VC1 encoder)...
Can you say more about this video?
Can you put some pictures from your source?

[benwaggoner]

Yes, sure for HDDVD but there is more and more exception with H264 (in UK, Germany, France,...)
and with BD, Sony prefer MPEG2 (bad implementation sometime) and H264, others H264 and some VC1.
We'll see how the numbers evolve, but so far we're the leading codec for HD optical - we're clearly very competitive in that market.


I hope you will not use Vista (lot of problem with driver and VC1 encoder)...
Can you say more about this video?
Can you put some pictures from your source?
I've been using Vista for many months for content creation without any issues. Adobe's CS3 Suite was the last public product I needed to be able to move happily to Vista 24x7.

The workstaiton has XP on it right now, but I think I'll put Vista Ultimate on it now just because you said that :).

[Golgot13]

The workstaiton has XP on it right now, but I think I'll put Vista Ultimate on it now just because you said that :).

The last VC1 encoder work well on Vista OS ?
Nice news to sell more new PC (my company don't want to change it) with better CPU
(there are now some nice motherboard which support well two Intel QuadCore)

[benwaggoner]

The last VC1 encoder work well on Vista OS ?
Yep, I'm running the new version on my Vista laptop without a hitch.

[Sharktooth]

a very slight OT:

http://pdl.warnerbros.com/wbol/uk/movies/300/300_F2_1080P.mov

http://gspsf.vo.llnwd.net/o10/300_Trailer2_8mbps_1080p29.98_5.1.wmv

which one looks better to you?

[davidcw]

Yes, sure for HDDVD but there is more and more exception with H264 (in UK, Germany, France,...)


It may be just coincidence but my 2 HD-DVD's out of 20 with visible artifacts were both encoded with AVC. Also, Planet Earth which was issued for the American market with an AVC encode didn't handle noisy shadow detail as well as the VC-1 encode and was given a lower picture rating by the reviewing site.
davidcw

[Dark Shikari]

It may be just coincidence but my 2 HD-DVD's out of 20 with visible artifacts were both encoded with AVC. Also, Planet Earth which was issued for the American market with an AVC encode didn't handle noisy shadow detail as well as the VC-1 encode and was given a lower picture rating by the reviewing site.
davidcwGenerally such artifacting is due to terrible VC-1/H.264 encoders rather than a problem with the standard itself. Theoretically, H.264 is superior to VC-1 in basically every way in terms of quality. Practically, its quite possible for H.264 to look worse in the case of a bad encoder.

[davidcw]

AVC encode didn't handle noisy shadow detail as well as the VC-1
davidcw

Could this be due to the in-loop filtering of AVC tending to remove more fine detail as it effects up to three pixels on either side off the boundary v's VC-1 effecting only one pixel either side of the boundary ?
davidcw

Edit: Here is the link to the review.

http://hddvd.highdefdigest.com/planetearth_us.html

[Dethis]

@ davidcw, about planet earth,

You (we) should not draw conclusions, about two encoder's capabilities, comparing one's 1080p24 vs other's 1080i30 encodings. Interlaced encoding at a 17% bigger framerate is a much more difficult task than 1080p24 (for any encoder).

[davidcw]

@ davidcw, about planet earth,

You (we) should not draw conclusions, about two encoder's capabilities, comparing one's 1080p24 vs other's 1080i30 encodings. Interlaced encoding at a 17% bigger framerate is a much more difficult task than 1080p24 (for any encoder).

Thanks for pointing this out. I had overlooked the fact that the AVC encode was 1080i30.
davidcw

[Golgot13]

Yep, I'm running the new version on my Vista laptop without a hitch.

Are you sure?
It is not slow like (speed to deliver) the video source file
that we wait since August (?)

If somebody can give video file without problem of copyright
to test HD encoder on film movie (with grain).
I really think (with problem of delay) that H264 is clearly better.

All electronic device will use H264 encoder, there were some nice device
( showed on CEATEC ).

[zambelli]

Thanks for pointing this out. I had overlooked the fact that the AVC encode was 1080i30.
davidcw
I think all VC-1 and AVC encoders used in HD/BD production thus far have been somewhat struggling with interlaced quality. Unlike back in the DVD days, HD encoding has been so focused on progressive content that interlaced encoding is getting left out of the spotlight. There's definitely more progress to be made there, though in my personal opinion, most 1080i content might do better being scaled and encoded as 720p60 content.

[zambelli]

a very slight OT:
http://pdl.warnerbros.com/wbol/uk/movies/300/300_F2_1080P.mov
http://gspsf.vo.llnwd.net/o10/300_Trailer2_8mbps_1080p29.98_5.1.wmv
which one looks better to you?
Were the sources the same?

[Sharktooth]

dunno about sources. the .mov source seems more grainy though. however they have almost the same bitrate but what im interested in is what one has less compression artifacts.
i already did a frame by frame comparison but i would like to have some feedback from others.

[Golgot13]

@Zambelli

No news from Ben for video source.

I hope we will have the source and VC1 encoding file (with PEP 1.07) soon.
Last development of H264 encoder are good (Mainconcept core,...)
and make better quality than PEP 1.06 (or VC1 SDK encoder).

[Sharktooth]

They dont provide the (in)famous source... they're not participating the MSU codec comparison too...
My guess is they're trying to take some time to "dodge" a smack...

[Golgot13]

My guess is they're trying to take some time to "dodge" a smack...

They will not dodge.... No, it's MS and the best team who made the famous VC1 codec.
It is :
" VC-1 achieves clearly superior quality to MPEG-2 at comparable bit rates, and has been judged superior to H.264 in several independent studies. "
Sorry Zambelli, I give another time the link:
http://www.microsoft.com/windows/windowsmedia/howto/articles/vc1techoverview.aspx#VC1ComparedtoOtherCodecs


I'm not really independant because I think VC1 can not make better quality than VC1
(since January when MS prohibited all french company to provide any info about PEP and VC1)
but all other people on this forum saw result made by others who are really independant (like Sagitaire, ...)

[zambelli]

They dont provide the (in)famous source... they're not participating the MSU codec comparison too...
My guess is they're trying to take some time to "dodge" a smack...
...Or the explanation could be something more benign, such as "Ben has been insanely busy, and the timing of MSU's codec comparison didn't really fit well into the MS codec team's schedule". But if you prefer to speculate, who am I to protest.

BTW, I don't think Ben ever said the source he was going to provide would favor VC-1 in any way, so I'm not sure how delaying the source delivery works in favor of MS VC-1.

(since January when MS prohibited all french company to provide any info about PEP and VC1
Where do I even begin? Microsoft can't forbid *everybody* to talk about something, especially about VC-1 which isn't even owned by Microsoft. Microsoft can sign NDAs (Non-Disclosure Agreements) with companies it licenses its software to, which is a fairly standard practice in the software world when pre-release or beta software is in question.

BTW, the text you keep linking to will be changed before the end of the month, removing some of the more ambiguous references to MPEG-2 and H.264. Though I'm sure you'll probably just keep linking to some archive page after that. :rolleyes:

[Sharktooth]

BTW, I don't think Ben ever said the source he was going to provide would favor VC-1 in any way, so I'm not sure how delaying the source delivery works in favor of MS VC-1.
Here's the dodge...

[Beastie Boy]

Why does every thread that tries to compare codecs or encoders end up becoming a stream of endless critisism aimed at Microsoft? They make a codec that you are free to use if you wish, but no-one has a gun to your head. If you don't like it and prefer to use an alternative, then move along and find something else to whine about.
When this thread started I was hoping it would be more about optimising codecs to get the best out of each one, and not reduced to a 'mine's better than yours' type of argument that is best suited to a playground.

I am certainly not a MS fanboy but I have a good deal of respect for both zambelli and Ben for continuing to post and offer help in these forums, despite the constant tarade of critism that they have to endure.

Rant over.

Cheers, Beastie.

[Sharktooth]

The criticism comes when M$ affirmed VC-1 is better than other codecs and that's clearly false (maybe it's only true for MPEG-2...). That's still on their website... Also in the same page they say VC-1 was designed to achieve state-of-the-art (in my dictionary it means the best solution) compressed video quality that may range from very low to very high bitrates. That's false to. Since actually both MPEG-4 AVC and MPEG-4 ASP (and by consequence also VP6 and VP7...) look better at low bitrates, while at high bitrates... well, you have the results here on this thread. So, yes, the criticism comes from the spreading of false informations so that criticism is TOTALLY appropriate.
The conclusion is M$ VC-1 is a mediocre codec. It stands between MPEG-2 and MPEG-4... maybe it will be further developed to something interesting, but the actual situation is MPEG-4 ASP is a preferable choice.

[Inventive Software]

@Sharktooth: In my SD challenge, I find that the VC-1 encodes I do are actually better than both Xvid and DivX, the ASP codecs using their respective insane settings. I'll do another encode this weekend with Xvid 1.1.3 and DivX 6.7, and you can compare the difference. ;)

[Sharktooth]

@inventive software: https://forum.doom9.org/showthread.php?p=1051937#post1051937 and https://forum.doom9.org/showthread.php?p=1054468#post1054468

[benwaggoner]

Sheesh, all this testiness!

As Alex said, things have been unexpecteldy and insanely busy for us both in the last few weeks. I've made some progress on the HD source clip, but it's not close to done - at least another two weeks, sorry. Rest assured, there are plenty of other projects I've had to put on the back burner as well.

I did render out a pretty interesting HD 10-bit motion graphics test if that would tide people over. It's mainly to test our 10-bit to 8-bit conversion, but if people want something to play around with, I can get it posted when I'm on campus next week. I warn you, though - I composed the soundtrack :).

As for all the issues about comparing VC-1 v. H.264, remeber that all we can do as codec nerds is compare implementations, not standards. We've now released our new VC-1 Encoder SDKs to software companies to incorporate in their products, so you'll be able to judge the progress we've made in the last year soon enough. FSDK11 wasn' the end of VC-1 optimization, any more than x264 was frozen in '06.

[Inventive Software]

Is that 10 bits per channel?