VC-1 formally approved by SMPTE!

[diogen]

...so in first time some HD DVD Video will be on DVD disc...It doesn't look like this will be the case. The first wave of HD-DVD (in US) will be VC-1 on DL 30GB disks. Sony on BD will be MPEG-2 on SL 25GB disks. It is not clear what titles and when will use H.264 AVC encoding.
You are right that some of the movies will have only ~8GB of video (<10Mbps bitrate), but thats just video.
...VC1 is available since January... This is true for the codec available to the studio mastering houses, not the public (as sid pointed out).

Amir on AVS claims the VC-1 encoding of the first US titles is so good (probably "The Last Samurai"), you can see dust in the air and people's spit while they are talking.

Diogen.

[diogen]

That codec sucks like hell. It depends on the source. T2EE shows it can be very good. This is not true for regular DVDs.
WVC1 should be much better. That's what everybody says...:sly:

Diogen.

[siddharthagandhi]

Everything Diogen said is right except about this:

"This is true for the codec available to the studio mastering houses,"

He was talking about the deprecated codec in Windows Media Encoder x64 edition. That codec is WMV 9 Advanced Profile, and is VC-1 compliant, but not fully compliant, and its worser than the "real deal" codec i described before.

[zambelli]

if you use VC1 at 16Mbps or MPEG2 HD, a movie of 90min will take 10,8GB of space only for video.
But in H264 at 8Mbps (I see a test of H264 with bitrate at 2,8Mbps HD DVD compliant in 1920x1080 of resolution),
the movie will take only 5,4GB and you can put it on DVD disc.
If their H.264 codec is more efficent than Microsoft's VC-1 codec, it's certainly not that much more efficient. The 16Mbps recommendation might be just that - a recommendation. How efficient the actual codec is depends on the implementation.

You really shouldn't take that info at face value. It's reasonable to expect that H.264 and VC-1 encodes of the same content will typically be of the same size/quality. Any efficiency advantage of one over the other will probably be due to the state of the particular codec implementation they're using, but I still doubt one codec will ever be more than 10% more efficient than the other.

VC1 is good codec because we don't need to have a big CPU to read HD video in this codec.
This is a very good point. VC-1 was designed to be a less complex codec than H.264, thus requiring less decoding resources. Any advantage you gain in encoding efficiency with H.264 over VC-1 - you're going to have to pay for it on the decoding side. (CABAC, for example.)

Golgot what you've seen since January isnt the REAL DEAL...if you read this thread. The real deal is the WVC1 Advanced Profile codec. That might even surpass AVC in terms of the best codec. What you've seen is WMVA, a (i quote) "deprecated codec". (dont wry i had to look that word up too :P) That codec sucks like hell. WVC1 should be much better.
Golgot still has a point - producing VC-1 AP bitstreams has been possible for everybody ever since the x64 Encoder came out. Regardless of what the actual FourCC code is, it's the bitstream that matters if you're looking to use VC-1 elementary streams in something like HD-DVD authoring.

BTW, now you've all got me questioning whether the x64 Encoder used the WVC1 FourCC or the old WMVA FourCC. I have to go find an x64 machine and find out for myself so it doesn't turn out I'm spreading misinformation. :)

Regardless of the FourCC being used, the important thing to understand about the 3 Advanced Profile Microsoft codecs is that they're just snapshots from a development timeline like any other.

x86 WMVA ----> x64 codec -----> x86 WVC1

The first one doesn't produce VC-1 AP compliant streams, the latter two do. From a dev standpoint, they all came from the same development branch, just from different points in time and different phases of development.
Even though the x64 encoder was released late in 2005, the codec itself was based on code dating to probably mid-2005.

[zambelli]

OK, guys, I screwed up. I just went and checked a x64 XP machine with WME9 x64 installed and it turns out the x64 WMV Advanced Profile codec actually uses the "WVC1" FourCC, not "WMVA" as I had previously said. Sorry about that!

This doesn't actually change much - the FourCC code itself is just an identifier to find the appropriate decoder. Everything else we discussed still holds true. The x64 AP codec is VC-1 compliant, but is probably about 9 months behind the current AP code.

I'll go back and edit the old posts just to make sure nobody new reading the thread gets confused.

[popper]

Why does VC-1 have to be encoded at 16 while AVC has to be encoded at 8 mbps? Why cant it be equal bitrate?

weather it has to be or not, i dont know, but its clear than the 8 Mbit (and lower) would be very usful for HD DTT broadcast on UKs FreeView and other countrys standards.

[siddharthagandhi]

"This doesn't actually change much - the FourCC code itself is just an identifier to find the appropriate decoder. Everything else we discussed still holds true. The x64 AP codec is VC-1 compliant, but is probably about 9 months behind the current AP code."

So the one that's coming out for beta will be a lot better than that one, right?

[IgorC]

Hi, zambelli
Can you provide a sample encoded by VC-1 codec. For example Harry Potter 2 Trailer http://multimediacom.free.fr/Download/
It is interesting to see how it performs at approx 500-600 kbps

[Doom9]

So where's the 32bit encoder so that we can face it off against x264? in2movies is just sheding a very bad light on Microsoft codecs...

[siddharthagandhi]

Zambelli said that the beta release is a codec, not an encoder.
I'd really like to see how it compares to Nero Digital, because ND has given me the best results and its one of the best AVC codecs as shown in MSU's and Doom9's codec comparisons.

[diogen]

So where's the 32bit encoder... It is supposed to be released this month, still two weeks left.
After checking MS site I noticed that the text of the announcement has changed:
what zambelli quoted in post #5 - April to be the release month - is not there anymore.
I do remember seeing the exact text as it is quoted in that post.
Does this mean the date has slipped or the release cancelled alltogether?
Can you comment, zambelli?

Diogen.

[IgorC]

I'd really like to see how it compares to Nero Digital, because ND has given me the best results and its one of the best AVC codecs as shown in MSU's and Doom9's codec comparisons.

Afaik. MSU and Doom9's comparisons compared High Profile Ateme AVC encoder that still not enable in Nero Recode2.
Nero Recode2 has only old Main Profile AVC. It's not bad and provide comparable quality as x264. Indeed last time x264 was improved siginifically - trellis, excelent ratetol, High Profile features like 8x8 dct and custom matrices, outstanding RD optimisations that makes x264 comparable only with close beta (not enabled yet) Ateme High Profile and Elecard.

[siddharthagandhi]

My results have shown that Nero Recode 2's codec is much better than x264. x264 doesn't even provide comparable quality to nero recode 2, in my opinion. But I don't want to start a codec debate.

[Golgot13]

NeroRecode use ateme encoder (in main profile). I use in my test for H264, the H264 software encoder
of Ateme (not directshow filter) : http://www.ateme.com/products/pcencoder.php

It is a good software and the quality of files encoded is one of better (1920x1080 24p at between 4Mbps to 5 Mbps
less I see difference with my 32" LCD from Samsung ).
There is some preset to be compatible with the hardware chipset (like STB7XXX from STMicroelectronic).
I do some test with Sigma Design chipset, it's wonderful...(but some options of H264 is not compatible
like GMC with old hardware chipset with Divx).

[Golgot13]

x264 is a best software to encode in H264 but the encoded file now will not compatible with
all hardware chipset present on market (because with can not change some options of H264).
So it's a good software to encode and watch the file on PC (with a good CPU).

But I think, in the near futur, some changement will do with x264 to be compatible
with hardware chipset.

And all new video hardware chipsets are directly compatible with H264 and VC1 (in High Definition). :D

[siddharthagandhi]

x264 doesn't give good results from fraps captures, whereas nero digital does...i'm doing some more testing on normal movies and videos

by the way how did u get a hold of ateme's high profile encoder? Did you buy it? Is there a demo?

[foxyshadis]

They had a closed beta a while back (been quite a while though! everyone thought recode would be updated as soon as it was over...), and I'm not sure, but it wouldn't surprise me if you could "obtain" it online.

[Golgot13]

I think the new NeroRecode will include this version of Ateme software. I see on show
(like CES, Ceatec, Cebit, IFA,...) that Nero is ready for HD DVD and BluRay (record disc software and
video software).
So for HD DVD, NeroRecode will be able to use High Profil for H264 and Advance Profil for VC1.
This version will be on market before september or this summer :D :D :D :D

[dohcmark8]

I hope the VC-1 codec is released soon.

I plan on compressing all my SD stuff with XViD and all my HD stuff with VC-1.

[siddharthagandhi]

Mad nice...Nero Recode will solve all my encoding needs

I currently use Nero Digital AVC and ASP for all my encodes, because for normal movies and Fraps captures it has give me the best results. Hopefully VC-1 will be as good as it sounds.

[trbarry]

I think the current WMV9 VCM is slipping a bit behind state of the art, making it compare less favorably with things like X264.

It would be nice to have a newer one.

- Tom

[zambelli]

So the one that's coming out for beta will be a lot better than that one, right?
Correct. Compared to both WMV9 MP (WMV3) and previous releases of WMV9 AP (WMVA and WVC1 x64), with new WVC1 codec you should see:

better quality at a given bitrate (CBR and 2pVBR)
smaller file size at given quality setting (1pVBR)

[zambelli]

So where's the 32bit encoder so that we can face it off against x264? in2movies is just sheding a very bad light on Microsoft codecs...
I haven't seen the in2movies encodes so I've no idea what they're doing. I'm assuming they're using Main Profile? Are they using CBR or VBR? 1pass or 2pass?

The beta of the new WVC1 codec will be out soon. You tested the alpha version back in December, so you know what to expect. You'll be happy to know that 2-pass VBR issues have been fixed and bitrate control is very accurate now.

[zambelli]

I do remember seeing the exact text as it is quoted in that post.
Does this mean the date has slipped or the release cancelled alltogether?
No, the plan is still to release it. The question is only which release vehicle to use. Keep in mind that Microsoft traditionally releases codecs with player/format updates. The only exception in recent years was the v9 VCM codec, so a lot of people are trying to figure out the appropriate release vehicle for this codec.

Sometimes I really envy companies that can just zip up the DLL together with a readme.txt and prop it up on a web server within minutes. ;) Here every release is like a Vista launch. :)

[stax76]

I haven't seen the in2movies encodes so I've no idea what they're doing. I'm assuming they're using Main Profile? Are they using CBR or VBR? 1pass or 2pass?

I don't think they state what they are using but wouldn't everything other than quality mody be flawed?

[siddharthagandhi]

"You tested the alpha version back in December, so you know what to expect."

Are you just talking to Diogen? Or was the alpha released to everyone?

[zambelli]

Can you provide a sample encoded by VC-1 codec. For example Harry Potter 2 Trailer. It is interesting to see how it performs at approx 500-600 kbps
720x576 at 500-600kbps? Umm, that's quite a stretch.

[Doom9]

Are you just talking to Diogen? Or was the alpha released to everyone?zambelli's reply was to me quote so guess whom he was talking to ;)

[Sagittaire]

720x576 at 500-600kbps? Umm, that's quite a stretch.

It's for this challenge
http://forum.doom9.org/showthread.php?t=90784

At this time x264 is the best by far
41.30 dB for 447 Kbps
43.90 dB for 886 Kbps

[bond]

720x576 at 500-600kbps? Umm, that's quite a stretch.thats the bitrate most people use around here (or at least the "pro" encoders seeing it as a sport to place a movie on 1cdr :D )

and i have yet to see wmv9 be able to beat even xvid at that bitrate (not to speak of x264)...

till i see this all the quality claims are not more than ...

[dohcmark8]

Question for the Microsoft guys... will we be able to encode in VC-1 using '1 Pass Quality VBR' mode?

As I see in the current VCM it doesnt let you choose a profile unles you use two pass, or am I missing something?

[CruNcher]

No, the plan is still to release it. The question is only which release vehicle to use. Keep in mind that Microsoft traditionally releases codecs with player/format updates. The only exception in recent years was the v9 VCM codec, so a lot of people are trying to figure out the appropriate release vehicle for this codec.

Sometimes I really envy companies that can just zip up the DLL together with a readme.txt and prop it up on a web server within minutes. ;) Here every release is like a Vista launch. :)

Would perfectly fit for the WMP 11 XP Release then ;) you just had to pre date it's release by 2 month, anyway "before end of june" could also mean today, tomorrow or next friday ;)

[diogen]

720x576 at 500-600kbps? Umm, that's quite a stretch.I'd more than happy if it can transparently encode 12Mbps satellite/cable hidef using about 5Mbps (fitting 2 hours on one DVD).
Do you know, zambelli, whether the 3-pass encoding option that Amir talked about on AVS (3rd pass for hand-tuning)
will also be available in the upcoming VC-1 release? Or is this only for studios?

Diogen.

[siddharthagandhi]

Whats a db measurement? Is that like a measurement of quality?

[foxyshadis]

PSNR (http://en.wikipedia.org/wiki/PSNR), a very rough measure of quality.

[zambelli]

thats the bitrate most people use around here (or at least the "pro" encoders seeing it as a sport to place a movie on 1cdr :D )
I understand the bitrate issue (trust me, I've been on this forum for a long time :)), but 720x576 is way too big for 500-600kbps to look good. However, after looking at Sagittaire's script, I see it's actually 720x304 at 900kbps that he requires. Much more reasonable, although still challenging.

and i have yet to see wmv9 be able to beat even xvid at that bitrate (not to speak of x264)...
till i see this all the quality claims are not more than ...
While I expect improvements in WVC1 over WMV3, I don't know how well it's going to compete with low-bitrate XviD or x264 encodes. The development focus in AP has been on a completely different end of the scale - HD encoding. Now, before everyone asks "What's the difference?" - I'm just saying that when one codec is optimized for 15Mbps HD encoding, and the other is optimized for 600kbps sub-SD encoding - comparing the codecs in this kind of test clearly gives the latter codec a home-court advantage. :)

But pictures speak more than words and tests will show us where WVC1 stands in that category. After all, that's what beta testing is all about.

[zambelli]

Question for the Microsoft guys... will we be able to encode in VC-1 using '1 Pass Quality VBR' mode?
As I see in the current VCM it doesnt let you choose a profile unles you use two pass, or am I missing something?
New WMV AP codec (WVC1) will support all encoding modes: 1p CBR, 2p CBR, 1p VBR, 2p VBR unconstrained and 2p VBR constrained.

[zambelli]

Do you know, zambelli, whether the 3-pass encoding option that Amir talked about on AVS (3rd pass for hand-tuning) will also be available in the upcoming VC-1 release? Or is this only for studios?
I remember briefly reading Amir's explanation and that stuff was news to me, too. I believe that's a property of the encoding tool that the studios are using, not the codec itself.

[siddharthagandhi]

I encode just about everything at 2 mbps, it would be nice if I could get it to 1.5 mbps or 1 mbps even. But would I see a benefit from VC-1 at 2 mbps? Thats not really HD bitrate, its not really an AVC bitrate either.

[IgorC]

Well, 500-600 kbps for 720x304 good clean source is enough ..... in case of x264.

There are some rumors from Vista and other MS applications beta testers:
New audiocodec WMA 10 is just tweaked WMA9. Not big changes. It's better than MP3 Lame at 128 kbit/s but it's still worse than itunes AAC. Info source hydrogenaudio

Other information that in the best case new VC-1 videocodec is just comparable to MPEG4 ASP.

Does anybody remember the promises of WMA9 codec?. It suposly was twice better than MP3 ( WMA9 64 = MP3 128). Practicly it's only slightly better.

[zambelli]

Well, 500-600 kbps for 720x304 good clean source is enough ..... in case of x264.
Enough for what? VHS quality? I'm not buying it. This is where I disagree with the whole 1CD approach to codec comparisons. It's soooo 1999 to try to encode movies at 600kbps and pretend they're acceptable quality. It's 2006 and we've got HD displays, we should be shifting our focus to better quality encoding, we don't have to settle for this "good enough" stuff anymore.
Even Doom9's shootout tests that involveed "Matrix: Revolutions" at 600kbps had the source resized down to 640x272 before encoding. Most bittorrent-distributed TV rips are 512x384 at at least 1000kbps.

There are some rumors from Vista and other MS applications beta testers: New audiocodec WMA 10 is just tweaked WMA9. Not big changes. It's better than MP3 Lame at 128 kbit/s but it's still worse than itunes AAC. Info source hydrogenaudio
Isn't LAME 3.96 just "tweaked LAME 3.91"? If the bitstream is locked, there is only so much improvement you can do to the encoder. Which codec did they evaluate? Standard, Voice or Pro?

[Liisachan]

Other information that in the best case new VC-1 videocodec is just comparable to MPEG4 ASP. If they are talking about the speed, that's possible. As for quality, today's WMV9 is already at least as good as MPEG-4 Part 2.

Does anybody remember the promises of WMA9 codec?. It suposly was twice better than MP3 ( WMA9 64 = MP3 128). Practicly it's only slightly better. You can not overestimate nor underestimate what does not yet exist. We will see...
And MPEG-related things could be like that too. I mean, for instance, it was claimed that HE-AACv2 48kbps was better than MP3 128, which is totally denied by the double-blind. It's normal that each company says their new product is the best and much better than existing things. Only the naive would believe that literally, but I guess VC-1 should be quite good.

[foxyshadis]

It's better than MP3 Lame at 128 kbit/s
Not meaning to pile on or anything, but that's like saying xvid is better than mpeg-1, or sorensen. mp3 has been outdated for many years, but it took a while for hardware (and apple) to show up and push more people to leave it behind. =p (Although lame has done a lot to push a great deal of extra complexity onto the encoding side to wring out extra quality.)

Does anybody remember the promises of WMA9 codec?. It suposly was twice better than MP3 ( WMA9 64 = MP3 128). Practicly it's only slightly better.
The funny part is that, compared to their optimization for high bitrate HD video, they're still optimizing the audio codec to the bottom of the bitrate end (which I don't disagree with, so is every other current audio format).

[GodofaGap]

There are some rumors from Vista and other MS applications beta testers:
New audiocodec WMA 10 is just tweaked WMA9. Not big changes. It's better than MP3 Lame at 128 kbit/s but it's still worse than itunes AAC. Info source hydrogenaudio

Actually this info is false: according to the latest multi-format listening test done by Sebastian Mares all codecs in the test (including WMA Pro, I-Tunes AAC and Lame MP3) were statistically tied at 128 kbps. So yes, MP3 performed just as good as I-Tunes and WMA.

http://www.hydrogenaudio.org/forums/index.php?showtopic=40607

Well, 500-600 kbps for 720x304 good clean source is enough ..... in case of x264.
I do not think that this is true in general. Perhaps for an almost perfect source.

[diogen]

...in the best case new VC-1 videocodec is just comparable to MPEG4 ASP...At 500kbps, 5Mbps, 15Mbps or across all bitrates?
If it's true for high bitrates, somebody should tell Warner and Universal about it since their HD-DVD encodes are all VC-1.

Diogen.

[siddharthagandhi]

"As for quality, today's WMV9 is already at least as good as MPEG-4 Part 2. "

Xvid is MPEG-4 Part 2...and you say that WMV9 is just as good as that? Pshhh....


Why does audio really matter? For internet distribution bitrates, at most its a difference of a couple of megabytes. I couldn't give a *blank* about audio codecs. Video is what I'm concerned about. And since it seems that VC-1 won't be better than AVC that means I'm going to have to stay with Nero Digital for the future.

[IgorC]

500-700 is preferible bitrate for most people. http://forum.doom9.org/showthread.php?t=101187

I actually do my backups at 700-1500 dependly from source complexity. But 600-700 is a good bitrate to benchmark codec performance.

Maybe VC-1 isn't so good. That's why there is no sample video. ;).

As it says all last comparisons (MSU, Doom9, Sagitarie etc. ) WMV9 ins't on par with MPEG-4 ASP.

Finally there is no any material proof (not even one trailer ) of that VC-1 will be as good as ASP.
That's why I asked to encode with VC-1 Harry Potter trailer.
Without video trailer all claims just are words and nothing more than simply words.

[siddharthagandhi]

We all just have to wait for the beta and then see if all the hype was worth it.

[HookedOnTV]

Enough for what? VHS quality? I'm not buying it. This is where I disagree with the whole 1CD approach to codec comparisons.

Here here. I don't get it either. What kind of display are people watching these on? 13"? What do you get by fitting it on a CD? Cost of CD's vs DVD's is a few pennies. 6/700k on a 50" display is horrible. All the darker areas have ringing and blocking. Does that not bother them? Drives me nuts.

Ok, so that's my rant... don't mean to offend anybody.

[dohcmark8]

Will there be a way for us to encode to VC-1 and retain the AC3 audio without using VCM in a AVI container?

What I mean is will Microsoft provide us with the tools to create HD-DVD/BR compliant streams (*.es)? or will a HD-DVD player be able to play VC-1 encoded in a AVI container with AC3 audio?