I just got my XA1 player, with The Last Samurai.

I also have the Japanese Wowow airing of the same movie. Japan HD is very high quality MPEG2. It' supposed to be around 22 Mbs, although it's more often around 20.

I compared the HD-DVD VC1 version with the DVHS one I have, using in both case an HDMI output, and a Qualia 004 as display on 4.5m wide screen.

Well, The Last Samurai on HD-DVD was disappointing. It still has some edge enhancement, and looks a bit less detailed and less 3D than the Wowow version.

The Wowow version might be a bit less clean though, maybe with a little more EE, but I like it better. Except for the Japanese subtitles!

Overall, both version are good and miles ahead any DVD, and much better than the low bitrate US airings. But I saw no definitive improvement with VC1, on the contrary.

I'd say that Sony might not be so wrong about choosing MPEG2 for their early title, considering that they will use even higher bit-rate and VBR coding instead of CBR.

Phantom of The Opera was completely artifact free, but although already quite good, did not have the ultimate details and 3D look that the best movies I have can show.

That's good news to me. MPEG2 has been around for decades and thus it's a mature technology. If VC-1 is close on it's maiden voyage into our homes then I expect it's potential means it usurps MPEG2 within the next 5 yrs.

I don't fault Sony for using MPEG2. It may be old but it's a known quantity and can look great.

Am I the only one who didnt notice EE with TLS HD DVD?? Watched it on my Infocus777 and 10 ft screen.


Kevin

Jediod

Probably not. EE seems to be like rainbows on a DLP. Some people are more sensitive than others.

Am I the only one who didnt notice EE with TLS HD DVD?? Watched it on my Infocus777 and 10 ft screen.

EE appears only on specific scenes, when there is a white background for example.

It's much less pronounced that the one you get on DVDs, but it's visible.

I just got my XA1 player, with The Last Samurai.

I also have the Japanese Wowow airing of the same movie. Japan HD is very high quality MPEG2. It' supposed to be around 22 Mbs, although it's more often around 20.

I compared the HD-DVD VC1 version with the DVHS one I have, using in both case an HDMI output, and a Qualia 004 as display on 4.5m wide screen.

Well, The Last Samurai on HD-DVD was disappointing. It still has some edge enhancement, and looks a bit less detailed and less 3D than the Wowow version.

The Wowow version might be a bit less clean though, maybe with a little more EE, but I like it better. Except for the Japanese subtitles!

Overall, both version are good and miles ahead any DVD, and much better than the low bitrate US airings. But I saw no definitive improvement with VC1, on the contrary.

I'd say that Sony might not be so wrong about choosing MPEG2 for their early title, considering that they will use even higher bit-rate and VBR coding instead of CBR.

Phantom of The Opera was completely artifact free, but although already quite good, did not have the ultimate details and 3D look that the best movies I have can show.

Good to hear differing opinions. Prepare to be flamed by the zealots though. How dare you conclude that VC1 is not superior!

Don't different regions create their own hidef masters for a movie? Aren't both taken from different sources? Is that the case here?

Well, The Last Samurai on HD-DVD was disappointing. It still has some edge enhancement, and looks a bit less detailed and less 3D than the Wowow version.

Have you calibrated the display for the Toshiba HD-DVD player ? Bland had some issues which he has solved. Take a look at this thread (http://www.avsforum.com/avs-vb/showthread.php?t=669185&page=1&pp=30) ...

Don't different regions create their own hidef masters for a movie? Aren't both taken from different sources? Is that the case here?

It could be of course, but the 2 transfers are pretty close. It's not like the difference between the D-Theater version of T2 and the WMV one.

Well, I might also wonder if it was the mastering / transfer process and not the codec itself.Could the title be better HDDVD transfer vs Wowow presentation.I know one thing you can get right up on the screen even in high contrast stuff and not see EE in Serenity. This was one thing I looked closely for ,edges looked like edges in film not video.

Art

Have you calibrated the display for the Toshiba HD-DVD player ? Bland had some issues which he has solved. Take a look at this thread (http://www.avsforum.com/avs-vb/showthread.php?t=669185&page=1&pp=30) ...

Yes, the first thing I noticed is that the black level was too high, but I corrected that. I got the same balance as with DVHS.

The 2 pictures are really close, the Wowow version being just a bit more snappy.

Well, The Last Samurai on HD-DVD was disappointing. It still has some edge enhancement, and looks a bit less detailed and less 3D than the Wowow version.

Well, I'm afraid there will be quite some upcoming HD-DVDs and BR with visible EE. It's simply in the masters. The Sony BR stuff from Sony masters of the last 5 years will pretty much all have visible EE.

Phantom of The Opera was completely artifact free, but although already quite good, did not have the ultimate details and 3D look that the best movies I have can show.
What are the best film sourced HD transfers you have (seen)?

I don't think this thread is supposed to proclaim that VC-1 < MPEG-2, but it does go to show that in the end, it doesn't really matter. MPEG-2 and VC-1 can look exactly the same quality.

Exactly, as long as there is enough bitrate.

He mentioned that the stuff commonly runs at 20Mbps. Is that CBR or VBR? What was Samurai encoded at?

It's broadcast, so it must be one pass CBR. There is not way to check accurately, the tape is 5C protected.

What are the best film sourced HD transfers you have (seen)?

From Wowow, Harry Potter 3 and Black Hawk Down (completely stunning in some low light scenes, not the grainy ones).

From BS/Hi, The Wrath of Khan.

From HBO, Drop Zone despite an average bitrate. And The Naked Gun 33 1/3.

I think that the initial photography and transfer are going to be the limiting factor, not the codecs.

could it be that DNR was used on the master for the Wowwow version? to remove film grain?

could it be that DNR was used on the master for the Wowwow version? to remove film grain?

The two transfers are too close for one to have DNR, and not the other one.

I just got my XA1 player, with The Last Samurai.

I also have the Japanese Wowow airing of the same movie. Japan HD is very high quality MPEG2. It' supposed to be around 22 Mbs, although it's more often around 20.

I compared the HD-DVD VC1 version with the DVHS one I have, using in both case an HDMI output, and a Qualia 004 as display on 4.5m wide screen.

Well, The Last Samurai on HD-DVD was disappointing. It still has some edge enhancement, and looks a bit less detailed and less 3D than the Wowow version.

The Wowow version might be a bit less clean though, maybe with a little more EE, but I like it better. Except for the Japanese subtitles!

Overall, both version are good and miles ahead any DVD, and much better than the low bitrate US airings. But I saw no definitive improvement with VC1, on the contrary.

I'd say that Sony might not be so wrong about choosing MPEG2 for their early title, considering that they will use even higher bit-rate and VBR coding instead of CBR.

Phantom of The Opera was completely artifact free, but although already quite good, did not have the ultimate details and 3D look that the best movies I have can show.

Thats because it is a Warner Title and I would have to say Apollo 13 doesn't look any better.

Well, The Last Samurai on HD-DVD was disappointing. It still has some edge enhancement, and looks a bit less detailed and less 3D than the Wowow version.

The Wowow version might be a bit less clean though, maybe with a little more EE, but I like it better. Except for the Japanese subtitles!

Overall, both version are good and miles ahead any DVD, and much better than the low bitrate US airings. But I saw no definitive improvement with VC1, on the contrary.
I am having a hard time integrating these three statements. In the first one, you say HD DVD is "disappointing." In the last one, you say it is "miles ahead [of] any DVD." Surely it can’t be both :).

In the second, you say WowWow version is "less clean" and "little more EE." The fact that you like it better seems to be a subjective preference as both of these are artifacts that should not be there. MPEG-2 distortion often shows up as ringing on the high frequency edges which you may be thinking are EE (high frequency components are quantized too much). And “less clean” comes from macroblocks which at high rate, probably come and go too fast for you to detect them as such, but would explain the "less clean" look. I am sure if you paused the video, you would see fair amount of macroblocking. Do the same with HD DVD (and other than reduced resolution due to field freeze), you won’t find the same problems.

At the end of the day, VC-1 does not increase the resolution, gamut, or contrast of the image. All it can do, is not add anything to the picture that was not there. So if you are seeing "less clean" video and "little more EE," then I say that VC-1 is doing its job in avoiding these problems. Of course, you might like the more artificial look of MPEG-2 video better. If so, then there is a fix for that: it is called the sharpness control on your display :).

Also note that TLS was an early title. Today, we can encode that at 14 Mbit/sec or even lower to achieve the same quality. It was one of those cases of the movie fitting in the allotted budget at that rate and folks moved on to the next title.

Finally, in studio tests, even 20-22 Mbit/sec VBR encoded MPEG-2 can not match the quality of VC-1 at much lower data rates. If this was not the case, there would be no pressure to add VC-1 to BD as there is.

Amir

Finally, in studio tests, even 20-22 Mbit/sec VBR encoded MPEG-2 can not match the quality of VC-1 at much lower data rates. If this was not the case, there would be no pressure to add VC-1 to BD as there is.Amir, just to correct this but VC-1 was added to the Blu-ray format a long time ago and recently the Scenarist Blu-ray authoring system (http://home.businesswire.com/portal/site/google/index.jsp?ndmViewId=news_view&newsId=20060424005395&newsLang=en) came out, which can encode and author in any of the three video codecs. As such it would make a lot more sense to use the past tense with your comment.

Finally, in studio tests, even 20-22 Mbit/sec VBR encoded MPEG-2 can not match the quality of VC-1 at much lower data rates. If this was not the case, there would be no pressure to add VC-1 to BD as there is.

Amir

Couldn't resist now could you... ;)

BD will be out soon enough and the real world comparison can begin then. If it looks good, it will, if it doesn't, you can crow the vitues of VC-1 to no end. And it's really a battle of encoders not codecs anyhow. Maybe Sony has a vastly better MPEG2 encoder then the one used in those studio tests? I doubt it's the same one Sony is using now, unless you can say for sure.

Who knows, someone might be working on H.264 encoder that might blow the doors off VC-1 by this time next year. :)

b2b

If this was not the case, there would be no pressure to add VC-1 to BD as there is. Bull. Regardless of reality, the "modern" codecs are perceived as new and better, even if that isn't true as a blanket statement. But that's where the pressure comes from. BD would of course want to be seen as looking towards the future.

Amir, just to correct this but VC-1 was added to the Blu-ray format a long time ago and recently the Scenarist Blu-ray authoring system (http://home.businesswire.com/portal/site/google/index.jsp?ndmViewId=news_view&newsId=20060424005395&newsLang=en) came out, which can encode and author in any of the three video codecs. As such it would make a lot more sense to use the past tense with your comment.
Unfortunately, there is no integration currently between our professional VC-1 encoder and Scenarist (their capability announced by them may be based on our standard VC-1 encoder in Windows). We very much hope such support can come in the future, but it is not there today.

The path to using VC-1 on BD would be to use our elementary streams created by our tool and mux that with the authoring tool used for BD discs. New authoring tools will take some time to get deployed by the studios but will be the right long term solution. For now, the format leaders bring this capability.

As a friendly note, it is so strange to have you keep telling me about the capabilities of our technology. I would be the first one to come here and declare support for VC-1 in BD titles. We are quite anxious to see that happen. As such, I don't need to be "corrected" with information gleaned from third-party press releases :).

Amir

Bull. Regardless of reality, the "modern" codecs are perceived as new and better, even if that isn't true as a blanket statement. But that's where the pressure comes from. BD would of course want to be seen as looking towards the future.
If so, why didn't they offer VC-1/AVC encoding in their authoring tools from day 1 like HD DVD did? Do they only want to look this way in press releases and not in reality?

Amir

I am having a hard time integrating these three statements. In the first one, you say HD DVD is "disappointing." In the last one, you say it is "miles ahead [of] any DVD." Surely it can’t be both :).

I stopped watching DVDs almost 3 years ago, they just look too bad.

Many Wowow (or HDnet) movies look stunning, but even some recent transfer like Wowow's Harry Potter 3 show some EE.

My disappointment is very relative. I was hoping to see EE gone on The Last Samurai, it's the first thing I looked for, and it was not.

I'm not bashing VC1 here. I'm sure that the EE was present in the master, and VC1 was certainly not going to remove it.

I'm thrilled by the HD-DVD and Blu Ray formats, the perspective of day and date HD releases with DVD ones is more than enough to make these 2 format worthwhile. And The Phantom of The Opera did not have *any* EE that I could see.

In the second, you say WowWow version is "less clean" and "little more EE." The fact that you like it better seems to be a subjective preference as both of these are artifacts that should not be there.

No, I don't think so. There was just a bit more texture on the Wowow version, it was not artificial at all.

MPEG-2 distortion often shows up as ringing on the high frequency edges which you may be thinking are EE (high frequency components are quantized too much). And “less clean” comes from macroblocks which at high rate, probably come and go too fast for you to detect them as such, but would explain the "less clean" look. I am sure if you paused the video, you would see fair amount of macroblocking. Do the same with HD DVD (and other than reduced resolution due to field freeze), you won’t find the same problems.

Although I haven't tried, I agree with you. The less clean look was certainly an MPEG2 artifact.

But both the "less clean" and "more texture" aspects were very subtle, the 2 transfers were very very close. My point is not about saying that one is bad and the other is good, just that the (hopefully limited in time) MPEG2 use by Sony is probably not going to be the disaster that some foresee.

If so, then there is a fix for that: it is called the sharpness control on your display :).

Nah, sharpness always gives ugly results.

Finally, in studio tests, even 20-22 Mbit/sec VBR encoded MPEG-2 can not match the quality of VC-1 at much lower data rates. If this was not the case, there would be no pressure to add VC-1 to BD as there is.

That was not the case for this particular transfer considering the global result.

2 titles is certainly way too low to form an opinion. With much more to come, I'm certainly hoping to see a few transfers that will exceed the best MPEG2 transfer I have. TLS and The Phantom were not one of them, that's all.

But again, I love the format and the perspective to not to have to go to extreme lengths to get viewable movies.

At the end of the day, VC-1 does not increase the resolution, gamut, or contrast of the image. All it can do, is not add anything to the picture that was not there.
Amir
Do you know for a fact that the Last Samurai master has no visible EE on it?

The path to using VC-1 on BD would be to use our elementary streams created by our tool and mux that with the authoring tool used for BD discs.Sure, most likely that will be the best way to do it since at least two of the six major studios already use Microsoft's VC-1 encoder. Still Amir you do know that this can be done with the Scenarist Blu-ray authoring tool?


As a friendly note, it is so strange to have you keep telling me about the capabilities of our technology.Well to be blunt Amir if you didn't keep implying that VC-1 can't be done on Blu-ray I wouldn't have to.


I would be the first one to come here and declare support for VC-1 in BD titles.I am not so sure about that, but I am sure you would admit to it after it has been officially announced by one or more of the studios.


We are quite anxious to see that happen. As such, I don't need to be "corrected" with information gleaned from third-party press releasesAmir, the Scenarist Blu-ray authoring tool can take the VC-1 encodings done by Microsoft's VC-1 encoder and author them for Blu-ray.

Don't different regions create their own hidef masters for a movie?

I'd hope not. Browse a few comparison pages at DVDbeaver and you'll see that most foriegn transfers use inferior sources compared to the version that was actually supervised with the DP or director. Fortunately this is becoming less of an issue as one transfer is used for all markets.

Amir, the Scenarist Blu-ray authoring tool can take the VC-1 encodings done by Microsoft's VC-1 encoder and author them for Blu-ray.

Richard, do you work at Sony's DVD center? Do you perhaps know someone that works there? What do you know about Sony's actual commercial authoring capability at this point in time?

I have a friend in DVD production, no not Sony, but the information he has within that community is that Sony is using MPEG-2 for initial release titles on BD not because they think that is the best codec, but because that is the only codec they are currently capable of authoring in. Also, at this moment, only Sony can author a Blu-Ray disc.

So, if my information is correct, and I have yet to see anything that would make me think it is not, then Amir is quite correct from a practical sense. It is one thing to say the Blu-Ray spec allows for VC-1, obviously, it does. However, theoreticals are not practicals. Theoretically, BD allows for VC-1. It just can't be done yet.

Unfortunately, there is no integration currently between our professional VC-1 encoder and Scenarist (their capability announced by them may be based on our standard VC-1 encoder in Windows)...

Who has access to the 'Pro VC-1 Encoder'?

Does Scenarist clearly state that they have the 'J6P VC-1 Encoder'?

Who has access to the 'Pro VC-1 Encoder'?
Currently we are offering it to studios/post production houses that have the ability to use it for HD DVD/BD production. The tool requires training and hand holding initially so it is not ready for general availability. Same is true btw to a larger extent in the case of the authoring tools themselves.

Does Scenarist clearly state that they have the 'J6P VC-1 Encoder'?
I don't know what it has in it. All I can tell you what it does NOT have in it (i.e. our pro encoder). For all I know, they could have written their own encoder from VC-1 spec, licensed it from someone else, or use the standard one in Windows. In the latter case at least, it is better than what you are implying. Our Pro encoder however, in addition to having better picture quality, also supports advanced features such as ability to use up to 8 CPUs to do parallel encoding and other workflow related features. These things are just as important as better picture quality.

Amir

Hopefully the Pro version shows up on UseNet soon.

Richard, do you work at Sony's DVD center? Do you perhaps know someone that works there? What do you know about Sony's actual commercial authoring capability at this point in time?Robert, you should never leap before you look. The Scenarist Blu-ray authoring system (http://home.businesswire.com/portal/site/google/index.jsp?ndmViewId=news_view&newsId=20060424005395&newsLang=en) is made by Sonic and was officially released a few days ago. It can handle VC-1 video streams and that would include the ones that are made with Microsoft's VC-1 encoder. Also here is one of the technical pages (http://www.sonic.com/products/Professional/Scenarist/specifications.aspx) that Sonic made for it.


Also, at this moment, only Sony can author a Blu-Ray disc.That would have been right a week ago. Today their are now two Blu-ray authoring systems and one of them can handle all three video codecs.


So, if my information is correct, and I have yet to see anything that would make me think it is not, then Amir is quite correct from a practical sense.Did you miss the part where Amir said "The path to using VC-1 on BD would be to use our elementary streams created by our tool and mux that with the authoring tool used for BD discs." Well now that their is a Blu-ray authoring system that can use VC-1 video streams that can be done.


Theoretically, BD allows for VC-1. It just can't be done yet.It can be done today :).

"It can be done today," you say.

Who has both the pro encoder and the authoring tool?

"It can be done today," you say.

Who has both the pro encoder and the authoring tool?

Many times Amir has said that MS would go anywhere, anytime to work with anybody that is willing to include VC-1 support for BD authoring. Sonic looks to be willing and they should be easy to find. ;)
I would be the first one to come here and declare support for VC-1 in BD titles. We are quite anxious to see that happen.
All that is left to do is for Amir to say... (in his best Captain Picard voice..) "Make it so.." :)

b2b

"It can be done today," you say.

Who has both the pro encoder and the authoring tool?Anyone who has, or gets, the VC-1 encoder from Microsoft and the Blu-ray authoring system from Sonic. Don't know which studios may end up doing that, but I can think of two studios that are very likely to.

Also note that TLS was an early title. Today, we can encode that at 14 Mbit/sec or even lower to achieve the same quality. It was one of those cases of the movie fitting in the allotted budget at that rate and folks moved on to the next title.


What do you mean by "it was an early title"? It was encoded some time ago?

That would seem to raise the question: Are some bashing Sony for choices made for authoring they started six months ago?

What is the lead time on a title authoring?

Gary

Many times Amir has said that MS would go anywhere, anytime to work with anybody that is willing to include VC-1 support for BD authoring. Sonic looks to be willing and they should be easy to find. ;)
Who said we are not working with them?

All that is left to do is for Amir to say... (in his best Captain Picard voice..) "Make it so.." :)

b2b
I wish one could write software that fast :). Or convince a post house to use a new tool overnight. My life would sure be a lot easier....

Amir

What do you mean by "it was an early title"? It was encoded some time ago?
This is a relative term. Relative to us making some of our recent encoder enhancements.

What is the lead time on a title authoring?
HD DVD systems started to get installed at some post houses nearly a year ago. A lot of training, workflow management, etc. has been put in place since then. So changing things doesn't happen overtime. Likewise, Sony has installed their system at various post houses. Note that studios probably did not have to pay for any of this due to strategic interest of the format leaders. All of this makes it harder for newcomers to get into the system.

These newer systems will most likely be used by smaller production houses first, before the majors. The majors are in the mode of cranking out titles, and not starting over with a new system. Once they have breathing room, maybe things change.

Amir

Who said we are not working with them?
Amir

Did they refuse?

Did they refuse?
Of course not.

Amir

Who said we are not working with them?

Ahh.. but if you don't come out and say it, some folks mistakenly assume the opposite..

I wish one could write software that fast :). Or convince a post house to use a new tool overnight. My life would sure be a lot easier....

Amir

Maybe a call from Warner Corp. offices to the post house, questioning the need for their future services might get their "convincing" process up to speed. ;)

b2b

Maybe a call from Warner Corp. offices to the post house, questioning the need for their future services might get their "convincing" process up to speed. ;)

b2b
Don't want to say who has done what. But why do you think they would demand a new system rather than having the existing one do the job? And who would pay for the new system?

Amir

Don't want to say who has done what. But why do you think they would demand a new system rather than having the existing one do the job? And who would pay for the new system?Amir, are you saying that Time Warner helped fund the Scenarist Blu-ray authoring system so they could use VC-1 with Blu-ray movies?

Amir, are you saying that Time Warner helped fund the Scenarist Blu-ray authoring system so they could use VC-1 with Blu-ray movies?
Not at all. How did you arrive at that conclusion?

Amir

Don't want to say who has done what. But why do you think they would demand a new system rather than having the existing one do the job? And who would pay for the new system?

Amir

I was under the impression that Warner was going with VC-1 for their BD releases. Without an authoring system, that would seem a difficult task to do, unless the existing one can do both BD & HD-DVD w/VC-1. Or that the Warner BD titles will be produced with MPEG2?

Actually Samurai would be a fine test to compare BD MPEG2 vs HD-DVD VC-1.. :)

b2b

Warner has made no announcements as to their codec choice for BD. People should not make any assumptions regarding that. Nor am I at liberty to say what they will do.

Amir

Not at all. How did you arrive at that conclusion?Well your post to b2bonez was a tad vague so I just took a stab in the dark to what it meant. Also Amir can't the Scenarist Blu-ray authoring system use VC-1 encodings that were made by Microsoft's VC-1 encoder? After all aren't they normal VC-1 compliant video streams?

Warner has made no announcements as to their codec choice for BD. People should not make any assumptions regarding that. Nor am I at liberty to say what they will do.

Amir

Well if the same titles end up on BD in MPEG2 and HD-DVD in VC-1, you better be getting your talking points ready.. That debate will go on forever.. ;)

b2b

Lots of odd questions about our work with Sonic here!

I just spent all week at NAB (16 hours of formal presentations, and about as much of booth duty - lost an octave of my voice!).

Sonic had a pod in the Microsoft booth demonstrating their VC-1 encoding and HD DVD authoring. We suppor them, and I'm enormously greatful to have a commerical, shipping product for both encoding and authoring I can point people towards.

Also, we announced our Windows Media Encoder Studio Edition. As of tomorrow (Friday), you'll be able to download this tool, which will enable you to make VC-1 in WMV with some of the same tools the studios use to make HD DVD. This will give folks a great chance to get a sense of what VC-1 is capable of.

The implementation of the codec isn't everything the studios have today (we're getting better every month), but certainly will be a big improvement for anyone who has played with our older implementation!

Well if the same titles end up on BD in MPEG2 and HD-DVD in VC-1, you better be getting your talking points ready.. That debate will go on forever.. ;)

The debate will be very, very short and decisive if the titles are using average data rates much below 25 Mbps...

Sure was fun demoing real, shipping transparent compression at the booth after all this time!

Oh, we got permission to quote that Serenity uses an average bitrate of 16 Mbps. That was another of the earlier titles that was finished, so we could certainly do it at a lower bitrate today, with some of the recent codec enhancements.

The debate will be very, very short and decisive if the titles are using average data rates much below 25 Mbps...

Sure was fun demoing real, shipping transparent compression at the booth after all this time!

Oh, we got permission to quote that Serenity uses an average bitrate of 16 Mbps. That was another of the earlier titles that was finished, so we could certainly do it at a lower bitrate today, with some of the recent codec enhancements.

I don't really see any advantage to doing the "bit rate limbo" unless you are working on a HD-DVD9 or the studio wants to save a few cents by using a SL15GB disc. If you have the space use it. Right now I would expect there is a lot more disk space available than there are hands to do all of that "hand tuning" to crunch the bits smaller.

Oh, BTW, are there any studios that are planning on doing encodes direct from DI masters (ie. skip the D5 tape step)? It would also be interesting to see some of Lowry's work on the old films coming from his pristine restored sources.

b2b

I don't really see any advantage to doing the "bit rate limbo" unless you are working on a HD-DVD9 or the studio wants to save a few cents by using a SL15GB disc. If you have the space use it. Right now I would expect there is a lot more disk space available than there are hands to do all of that "hand tuning" to crunch the bits smaller.
Sure. There are title where they hit their bit-budget, and called it done, even if they could have done it lower with some more reencoding. But more efficient encoding offers more flexibility, and is always welcome to have as an option.

Oh, BTW, are there any studios that are planning on doing encodes direct from DI masters (ie. skip the D5 tape step)? It would also be interesting to see some of Lowry's work on the old films coming from his pristine restored sources.
I of course can't speak to what particluar studios are doing. I can say for my part that I've done a few DI to VC-1 encodes, and they looked spectacular. Now that NAB is over, I'll try to figure out a good way to post them.

Sure. There are title where they hit their bit-budget, and called it done, even if they could have done it lower with some more reencoding. But more efficient encoding offers more flexibility, and is always welcome to have as an option.
No doubt, flexibility is good.
I of course can't speak to what particluar studios are doing. I can say for my part that I've done a few DI to VC-1 encodes, and they looked spectacular. Now that NAB is over, I'll try to figure out a good way to post them.
No identities needed, just wondering if there were any in the works for release products. It just looks like D5 might be a limiting factor in the never ending quest for PQ. :)

b2b

Sure was fun demoing real, shipping transparent compression at the booth after all this time!Ben, do you think Serenity is a transparent compression?

Oh, we got permission to quote that Serenity uses an average bitrate of 16 Mbps. That was another of the earlier titles that was finished, so we could certainly do it at a lower bitrate today, with some of the recent codec enhancements.So, does the latest VC-1 implementation solve some compression artifacts, especially for pulsing issue, of Serenity at even lower bit rate?

Do you know why the post house who is doing Warner titles seems to use DNR to wipe out film grains and small textures? Or is it those master's characteristic?

I don't really see any advantage to doing the "bit rate limbo" unless you are working on a HD-DVD9 or the studio wants to save a few cents by using a SL15GB disc. If you have the space use it.To add fancy stuff, like PiP, they need to reduce max/average bit rate and capacity of the main video.

I of course can't speak to what particluar studios are doing. I can say for my part that I've done a few DI to VC-1 encodes, and they looked spectacular.

Do these encoders take 10 bit 4:4:4 input and convert to 8 bit 4:2:0 internally before compression? If not, what hardware/software does this?

B
Do you know why the post house who is doing Warner titles seems to use DNR to wipe out film grains and small textures? Or is it those master's characteristic?

How do you figure that happened? Do you see artifacts?

No identities needed, just wondering if there were any in the works for release products. It just looks like D5 might be a limiting factor in the never ending quest for PQ. :)
D5 issues can slow down the process certainly. For that first HD MPEG-2 title I did back in 2003, there were D5 "hits" - green square all over the opening credits. A recapture from a better deck fixed most of it, and a lovely few days rotoscoping in After Effects did the rest.

I cant' tell you how awesome it is for a D5-HD master tape (10-bit 4:2:2, 8:1 compression) being the potential limit to how good a consumer disc format is :).

With VC-1, the primary limitaiton is the quality of the master at this point. A bad telecine would be a big issue than a bad D5 - D5 never gets THAT bad from a compression perspective, compared to a poor source.

Do these encoders take 10 bit 4:4:4 input and convert to 8 bit 4:2:0 internally before compression? If not, what hardware/software does this?
Yes, there needs to be a conversion from 10-bit to 8-bit, and from whatever color depth to 4:2:0. There are different places this can happen (in the DDR, with a utility we made, inside the encoder in the case of Sonic). Doing that well, especially getting the optimum dither from 10-bit to 8-bit, can definitely make a difference.

FYI, the new Windows Media Encoder Studio Edition Beta that'll be available for download sometime today will include the advanced scaling and dithering modes we provide to the studios, so you can give them a shot yourself.

To add fancy stuff, like PiP, they need to reduce max/average bit rate and capacity of the main video.
A little bit. The PIP content we've done so far typically hasn't been particularly bit-hungry. 4 Mbps would be the highest peak for PIP, and most PIP wouldn't use that much.

Ben, do you think Serenity is a transparent compression?
I haven't seen Serenity all the way through, personally - that was finished before my first trip down to the studios. People I've talked to about it seem quite pleased.

I have seen "Last Samurai" and "Phantom of the Opera" all the way through, and I consider them transparent to the source.

So, does the latest VC-1 implementation solve some compression artifacts, especially for pulsing issue, of Serenity at even lower bit rate?I've been out of the loop on news for a couple of weeks due to the "NAB bubble" - do you have a link to a description of that?
Do you know why the post house who is doing Warner titles seems to use DNR to wipe out film grains and small textures? Or is it those master's characteristic?
I don't know of any post houses or studios that use DNR as a matter of course - it's be a case by case basis. Again, I can't get into specifics of title or particular studios, but post-D5 DNR looks to be a relative rarity for VC-1 titles. I expect it'll be mainly the domain of BD MPEG-2 titles, where DNR of some sort (preprocessing or in codec) would be needed to avoid blocking/ringing for longer titles.

How do you figure that happened? Do you see artifacts?So far, I saw TLS, POTP and Million Dollar Baby. Commonly seen characteristics of those three titles are, slightly soft image, less film grain at brighter area and little bit un-natural noise movement at flat area. For example, in one scene of MDB around 00:07:20, there is a long shot for a street. When you look at the brick wall of a building on the right hand side, you'll notice that while camera frame is slightly moving (vibrating), the brick wall movement does not follow the camera frame movement, it looks moving independently. This is a typical artifact of the temporal DNR.

I have seen "Last Samurai" and "Phantom of the Opera" all the way through, and I consider them transparent to the source.
I've been out of the loop on news for a couple of weeks due to the "NAB bubble" - do you have a link to a description of that?For Last Samurai and Phantom, I felt those looked slightly softer than I expected. robena felt same thing. I thought my feeling came from pre filtering to reduce film grains to make it look more "cleaner" than the original master and make encoding process easier. Since Wowow content is a broadcast thing, I suspect that they don't use pre filtering by content basis, just feed D5 master to the encoder.

For Serenity, you'll get pulsing issue (you know well what pulsing means, I think) all the time at flat area. Slow motion playback makes you easier to recognize that. At bright area, film grains produce compression artifacts, it does not look like natural grains. I don't recall exact position but some explosion scene of a space ship, I found obvious block noises that I can't find on TLS or POTO.

Again, I can't get into specifics of title or particular studios, but post-D5 DNR looks to be a relative rarity for VC-1 titles. I expect it'll be mainly the domain of BD MPEG-2 titles, where DNR of some sort (preprocessing or in codec) would be needed to avoid blocking/ringing for longer titles.I think it belongs to the post house's decision. Whether post houses use DNR or not would vary title by title basis. If there is too much film grains, some post house will try to keep them as much as possible, with many tweakings and hard time jobs, some post house will use DNR to wipe out those noises for ease of encoding tasks.

Regarding softness, was it ever confirmed that the Toshibas aren't pre-filtering the interlaced outputs like the HD-DVD specs seemed to indicate as a requirement? Now that people have players, are there any user selectable settings for enabling or disabling that feature?

Regarding softness, was it ever confirmed that the Toshibas aren't pre-filtering the interlaced outputs like the HD-DVD specs seemed to indicate as a requirement? Now that people have players, are there any user selectable settings for enabling or disabling that feature?

No, no user selectable feature concerning filtering.

I wonder about that too. Despite Amir reassurances, the fact that TLS was not sharper than the Wowow airing makes me think that these interlaced output might be filtered.

We will only know for sure when the same title will be available on Blu Ray and HD-DVD. If the Blu Ray version at 1080p/24 is much sharper, that will be a definitive proof.

So far, I saw TLS, POTP and Million Dollar Baby. Commonly seen characteristics of those three titles are, slightly soft image, less film grain at brighter area and little bit un-natural noise movement at flat area. For example, in one scene of MDB around 00:07:20, there is a long shot for a street. When you look at the brick wall of a building on the right hand side, you'll notice that while camera frame is slightly moving (vibrating), the brick wall movement does not follow the camera frame movement, it looks moving independently. This is a typical artifact of the temporal DNR.

Oh no! I'm very sensitive to DNR artifacts. Titles that have anything not marginal I will boycott/rent only. I have had enough of this **** on DVD and lately 35mm prints from DIs. :mad: :mad: :mad: :mad: :mad:

We will only know for sure when the same title will be available on Blu Ray and HD-DVD. If the Blu Ray version at 1080p/24 is much sharper, that will be a definitive proof.

As long as the codec is the same. 2G HD-DVD players will hopefully have 1080p outputs and that would be a better test. I am very interested to see how Sony's implementation of MPEG-2 fairs against VC-1 or MPEG-4 in an actual HiDef DVD test. If Bluray outperforms HD-DVD, I think Bluray will pick up a lot of support from AVSers. If they fall short, there may be more defectors to HD-DVD.

Some may jump ship, some may get both. Most won't. So far, the only studio we know for sure will use MPEG2 is Sony. It is probably fairly safe to assume that Warner will be releasing their Blu-ray titles using the same codec as for HD DVD. (Then again, from an excellent* Segal movie: Assumption is the mother of all ****-ups...)

I won't jump ship. It is hard already, and that only with a handful of HD DVD titles out. But, with my best french accent, I shall resist. I'm in this for the long run, and won't judge the platform from a few early releases. (But I am not greatly worried. Heads will roll at Sony should they release sub-par titles, when they KNOW what they are competing against. I'm sure they'll come through with some excellent PQ on their releases.)

I just wish I could get Blu-ray NOW! It is going to be a loooong summer (and autumn). But, some great movies coming out. Can't wait to get them in full HD glory at home. Anyone done a HD DVD vs. Blu-ray studio thingy on the upcoming summer blockbusters?

*) Definition of excellent slightly less excellent than Websters definition...

Yes, there needs to be a conversion from 10-bit to 8-bit, and from whatever color depth to 4:2:0. There are different places this can happen (in the DDR, with a utility we made, inside the encoder in the case of Sonic). Doing that well, especially getting the optimum dither from 10-bit to 8-bit, can definitely make a difference.

FYI, the new Windows Media Encoder Studio Edition Beta that'll be available for download sometime today will include the advanced scaling and dithering modes we provide to the studios, so you can give them a shot yourself.

I am glad that you guys are thinking hard about proper dithering.

When I play HD content on my PC, I am stuck in the RGB 8-bit per color world... Color banding is an issue with HD playback unless the content is pre-dithered well. This shows up most prominently on outdoor scenes with sky and clouds that have subtle color variation.

I watched Serenity. That was a big WOW. It's what I was hoping for, an image (almost) totally devoid of any artefacts and EE. Much better than TLS.

There was just a 2 minutes scene near the ending where I saw some EE, they probably were a bit sloppy at that point.

The image dynamic was astounding. It was subjectively extremely bright and vivid.

I hope that either Blu Ray or HD-DVD will keep their word about day and date with DVD releases. We're finally here!

Oh no! I'm very sensitive to DNR artifacts. Titles that have anything not marginal I will boycott/rent only. I have had enough of this **** on DVD and lately 35mm prints from DIs. :mad: :mad: :mad: :mad: :mad:
No DVNR was used on those titles. Please go and see for yourself. You don't have to take word of a BD insider when the real thing is here ;). Or, read the reviews in the HD DVD Player section and see the ton of kudos on picture quality. Not one matches what Rio has said. Not even one!

Amir

amirm,

I am still very puzzled by this.

It appears to be your statement that VC-1 is better than MPEG2 under certain reasonable assumptions. It is also clear that VC-1 is a part of the Blu-Ray spec as well as HD DVD. Finally, Microsoft has announced that they are providing considerable help to studios producing HD DVD disks.

Questions: Is Microsoft also helping Blu-Ray? If not, why not? Were you not asked? Did you offer? It hardly seems reasonable for Microsoft to tilt so strongly to one of its own customers, unless of course Sony didn't want the help.

If Sony didn't want the assistance, is it possible that their MPEG2 encodings will be as good or better than VC-1 (which Sony has suggested to be the case)?

amirm,

Questions: Is Microsoft also helping Blu-Ray? If not, why not? Were you not asked? Did you offer? It hardly seems reasonable for Microsoft to tilt so strongly to one of its own customers, unless of course Sony didn't want the help.

This has been asked and answered a thousand times.
Sony didn't want the help.

I am surprised that no one has taken robena's post to pieces! :) Generally when something like this is posted, VC-1 camp comes up with very good defense.

I (and at least one other Blu-ray supporter) criticized Sony in the 'Main' thread for their choice of MPEG2 -- I still don't like their choice -- but I am now thinking, probably Sony is right in their choice, at least for the time being. I know VC-1 and H.264 AVC have much more head room to grow, and completely knock everyone off their feet, but it seems to me that in their present state, they are not providing tangible quality advantage over MPEG2 VBR (other than disc space which is a big plus, for sure). A couple of more months to go before I get my answer.

From what I have seen so far, it seems that Sony may get away with their choice (MPEG2) for the 'foreseeable future'. When will VC-1 encoded titles reach the quality that MPEG2 VBR can not touch? Will the next batch of titles provide that edge?

Has anyone on this forum played around with MPEG2 VBR encoding for high-def?

Ben/Amir, I am not bashing VC-1, to be clear. I am just disappointed that MPEG2 CBR is so close in terms of picture quality of these released titles. I have high expectations from VC-1 (and H.264 AVC).
Are we there yet? Are we there yet? Are we there yet? :)

I know VC-1 and H.264 AVC have much more head room to grow, and completely knock everyone off their feet, but it seems to me that in their present state, they are not providing tangible quality advantage over MPEG2 VBR (other than disc space which is a big plus, for sure). A couple of more months to go before I get my answer.


Uhh... added disc space is all they are supposed to provide. A codec's job is not to make a source look better than another codec, but to do it in less space. VC-1 has done that job admirably.

The goal is perceived transparancy, and MPEG-2 and VC-1/H.264 can do it, with the latter two hitting it at a significantly lower bitrate.

This has been asked and answered a thousand times.
Sony didn't want the help.

So I understand, and therefore I would expect that Sony would produce some excellent encodings using the MPEG2 technology.

Altogether I find the situation very puzzling, but we will see what the results are in another two months.

Uhh... added disc space is all they are supposed to provide. A codec's job is not to make a source look better than another codec, but to do it in less space.


Well, yeah. However, '...make a source look better...' is not what I am arguing about.


VC-1 has done that job admirably.

16Mbps is only marginally better. I have much higher expectation from VC-1. I am sure it will get there, though.


The goal is perceived transparancy, and MPEG-2 and VC-1/H.264 can do it, with the latter two hitting it at a significantly lower bitrate.

Exactly. However, 16Mbps (Serenity) isn't significantly superior to 18Mbps (which apparently Sony plans), does it? VC-1 camp has been saying that the bit-rates that Sony plans to use will not give transparency. Now, from this thread, it looks like 20-22Mbps MPEG2 CBR is giving close to transparency (with some minor artifacts). My extrapolation is that 18Mbps VBR may make it transparent. Do you think it will not? It will also be interesting to see how 16Mbps MPEG2 VBR looks (just out of curiosity).

The context here (my previous post) is Sony's choice of MPEG2. If Sony's MPEG2 encoded titles look as good or better than VC-1, then their choice is justified considering that they have real-time encoder. As far as space is concerned, I will not cry if that means less extras! :)

There has been a huge debate regarding MPEG2 quality (EE, macro-blocks, etc) Vs advanced codecs. Real world example here seems to prove it is not that big of a difference. In future, I am 100% sure advanced codecs will far surpass what MPEG2 can achieve, but I hope by then Sony will make a move towards them too (or that 50GB will become ubiquitous).

I want to see advanced codecs on Sony's titles as much as the next AVSer, but I will no longer crucify them because of their choice of MPEG2 for now. Of course in less than two months, we will know for sure.

I am surprised that no one has taken robena's post to pieces! :) Generally when something like this is posted, VC-1 camp comes up with very good defense.

I (and at least one other Blu-ray supporter) criticized Sony in the 'Main' thread for their choice of MPEG2 -- I still don't like their choice -- but I am now thinking, probably Sony is right in their choice, at least for the time being. I know VC-1 and H.264 AVC have much more head room to grow, and completely knock everyone off their feet, but it seems to me that in their present state, they are not providing tangible quality advantage over MPEG2 VBR (other than disc space which is a big plus, for sure). A couple of more months to go before I get my answer.

From what I have seen so far, it seems that Sony may get away with their choice (MPEG2) for the 'foreseeable future'. When will VC-1 encoded titles reach the quality that MPEG2 VBR can not touch? Will the next batch of titles provide that edge?

Has anyone on this forum played around with MPEG2 VBR encoding for high-def?

Ben/Amir, I am not bashing VC-1, to be clear. I am just disappointed that MPEG2 CBR is so close in terms of picture quality of these released titles. I have high expectations from VC-1 (and H.264 AVC).
Are we there yet? Are we there yet? Are we there yet? :)

It should be noted that this isn't a "battle of the codecs", but a battle of the encoders. Sony found a MPEG2 encoder solution that met their quality and usability needs for a professional production environment. MS has provided their VC-1 encoder for the first HD-DVD titles.

I suspect that Sony, faced with rolling out the completely new physical spec of BD-ROM, felt more comfortable with the mature MPEG2 to create HD titles. Not having that challenge, HD-DVD producers opted to take their "new technology risk" with VC-1 and the results have been fine.

What will be interesting is to objectively (as best as possible) compare the results for MPEG2 vs VC-1. I think what we will find, except for the most ardent of nitpickers, is that both will be more than acceptable.

b2b

I am surprised that no one has taken robena's post to pieces! :) Generally when something like this is posted, VC-1 camp comes up with very good defense.

I (and at least one other Blu-ray supporter) criticized Sony in the 'Main' thread for their choice of MPEG2 -- I still don't like their choice -- but I am now thinking, probably Sony is right in their choice, at least for the time being. I know VC-1 and H.264 AVC have much more head room to grow, and completely knock everyone off their feet, but it seems to me that in their present state, they are not providing tangible quality advantage over MPEG2 VBR (other than disc space which is a big plus, for sure). A couple of more months to go before I get my answer.

From what I have seen so far, it seems that Sony may get away with their choice (MPEG2) for the 'foreseeable future'. When will VC-1 encoded titles reach the quality that MPEG2 VBR can not touch? Will the next batch of titles provide that edge?

Has anyone on this forum played around with MPEG2 VBR encoding for high-def?

Ben/Amir, I am not bashing VC-1, to be clear. I am just disappointed that MPEG2 CBR is so close in terms of picture quality of these released titles. I have high expectations from VC-1 (and H.264 AVC).
Are we there yet? Are we there yet? Are we there yet? :)Given a high enough bit rate, any of these codecs can achieve transparency. The problem with Sony using MPEG2 is that it requires a higher bitrate, thus eating into BD's capacity advantage (not to mention that most BD discs will be only 25GB compared to most HD-DVD discs being 30). In order for BD to achieve its greatest potential, it will have to abandon MPEG2 in favor of the more modern codecs. As it stands right now, the typical BD disc has no real world technical advantage over the typical HD-DVD disc, in fact, the opposite is the case.

So I understand, and therefore I would expect that Sony would produce some excellent encodings using the MPEG2 technology.

Altogether I find the situation very puzzling, but we will see what the results are in another two months.

Puzzling indeed, and perhaps you may be wondering the same as I - if you have someone who is trying to be your lapdog as Microsoft would like to imply that they have done for Sony (in "offering" so much help), wouldn't you just accept?

One would wonder if the conditions they have for Sony is ultimately inferior to what they might have HD-DVD. The consequence of not accepting obviously is to have that fact "advertised" aggressively here.

Given a high enough bit rate, any of these codecs can achieve transparency. The problem with Sony using MPEG2 is that it requires a higher bitrate, thus eating into BD's capacity advantage (not to mention that most BD discs will be only 25GB compared to most HD-DVD discs being 30). In order for BD to achieve its greatest potential, it will have to abandon MPEG2 in favor of the more modern codecs. As it stands right now, the typical BD disc has no real world technical advantage over the typical HD-DVD disc, in fact, the opposite is the case.
This is all incorrect, as BD50 titles have been announced.

It should be noted that this isn't a "battle of the codecs", but a battle of the encoders.

Huh? That makes no sense. If you think Sony has discovered some magic bullet MPEG-2 encoder, you are in for an unpleasant surprise. MPEG-2 will always be MPEG-2. Newer encoders may refine it to some degree, but there is no way in hell any HD (1080p) MPEG-2 encoding at 18 mb/s will ever look as good as the same thing encoded with VC-1 at 14 mb/s.

However, 16Mbps (Serenity) isn't significantly superior to 18Mbps (which apparently Sony plans), does it?

On what is this statement based? Certainly not real-world experience. BD isn't out. Sony hasn't demo'd BD at anything like the bit rates being discussed in this thread. The only examples of hi-def MPEG-2 at bit rates below 20 mb/s are OTA broadcast, HD cable, and HD satellite. None of these examples can carry HD-DVD's jock strap with regards to image and audio quality. The only MPEG-2 HD material that compares favorably to VC-1 on HD-DVD is D-Theater, and that is at significantly higher bit rate that 18 mb/s, and even then is only comparable or below in quality.

This is all incorrect, as BD50 titles have been announced.

Yeah, right. And when you have one and a player to play it on, you will have something to add to this discussion.

I suspect that Sony, faced with rolling out the completely new physical spec of BD-ROM, felt more comfortable with the mature MPEG2 to create HD titles.I would be shocked if meetings and information that went into Sony's decision didn't include mention of how well they have done with Superbit at selling the same movie more than once, or basically the opportunity to sell movies with MPEG2 and then later on sell the same movie with an advanced codec (or DL). Transparency to the master leaves less room for reselling later and I think this is one area where I think the format war is a good thing (giving each side an incentive to not hold back too much for later).

--Darin

This is all incorrect, as BD50 titles have been announced.
But they've been announcing it for awhile now. The question is, can they reliable and profitably produce BD50 ROM?... That is a question that will be answered positively when the start actually doing it. I think most people assume BD50 is not something that will be ready from the "get go", but will be introduced later on.

Huh? That makes no sense. If you think Sony has discovered some magic bullet MPEG-2 encoder, you are in for an unpleasant surprise. MPEG-2 will always be MPEG-2. Newer encoders may refine it to some degree, but there is no way in hell any HD (1080p) MPEG-2 encoding at 18 mb/s will ever look as good as the same thing encoded with VC-1 at 14 mb/s.


What about what robena has observed? 20/22Mbps. It is not even VBR! I haven't seen Sony's encodings, I can extrapolate, though.

I agree with b2bonez. At least for short term, it will depend on the quality of the encoder, not the underlying codec (implementation Vs specification). If Sony does have good MPEG2 encoder (I don't think any magic is required; good engineering should be fine), basically in the end we will have to see the movies to make the final judgement.

It is ironic that you are judging un-released MPEG2 encodings already while ignoring what the original poster observed! :rolleyes:

On what is this statement based? Certainly not real-world experience. BD isn't out.
I am talking about bitrates. Serenity, we know. For Sony I assumed. Not much to differentiate based on current titles.

None of these examples can carry HD-DVD's jock strap with regards to image and audio quality.

If you want to ignore this thread, please do; I can't. I am here to learn.


PS: Please mention names of posters when you quote.

But they've been announcing it for awhile now. The question is, can they reliable and profitably produce BD50 ROM?... That is a question that will be answered positively when the start actually doing it

Those are good questions.


I think most people assume BD50 is not something that will be ready from the "get go", but will be introduced later on.
That is my understanding as well.

On what is this statement based? Certainly not real-world experience. BD isn't out.

Again... What will be interesting is to objectively (as best as possible) compare the results for MPEG2 vs VC-1. I think what we will find, except for the most ardent of nitpickers, is that both will be more than acceptable.

I can only guess that you getting practise in.. :rolleyes:

b2b

I would be shocked if meetings and information that went into Sony's decision didn't include mention of how well they have done with Superbit at selling the same movie more than once, or basically the opportunity to sell movies with MPEG2 and then later on sell the same movie with an advanced codec (or DL). Transparency to the master leaves less room for reselling later and I think this is one area where I think the format war is a good thing (giving each side an incentive to not hold back too much for later).

--Darin
Whatever their incentive, it does not prevent them from trying to achieve "superbit" with MPEG2 should the need arise (whether it is because of competition or for reselling).

I would be shocked if meetings and information that went into Sony's decision didn't include mention of how well they have done with Superbit at selling the same movie more than once, or basically the opportunity to sell movies with MPEG2 and then later on sell the same movie with an advanced codec (or DL). Transparency to the master leaves less room for reselling later and I think this is one area where I think the format war is a good thing (giving each side an incentive to not hold back too much for later).

--Darin

With DVD that was a "good" (business wise) strategy. However with the format war going on, such a strategy caries with it a significant amount of risk. By most/all accounts HD DVD (or studios producing movies on HD DVD) is providing transparency out of the gate. What happens if Blu-ray doesn't match that? Being that these will be videophile formats, limited to the high end for the near term, anything less than transparency (on either side), when the competition is providing it, will be torn to shreds in the reviews, both "professional" and amature.

What about what robena has observed? 20/22Mbps. It is not even VBR! I haven't seen Sony's encodings, I can extrapolate, though.

What about what robena has observed? Who is robena and what HAS he observed? Some Japanese broadcast that no one else can verify? Is that the best you can base your conclusions on?

I agree with b2bonez. At least for short term, it will depend on the quality of the encoder, not the underlying codec (implementation Vs specification). If Sony does have good MPEG2 encoder (I don't think any magic is required; good engineering should be fine), basically in the end we will have to see the movies to make the final judgement.

Assumptions based on ignorance supported by irrational bias. Again, I am at a loss.

It is ironic that you are judging un-released MPEG2 encodings already while ignoring what the original poster observed!

I couldn't care less what the original poster observed. Until there is evidence that others, including me, can observe and verify, there is nothing. That is what some of you in this thread are basing your assumptions and opinions on, NOTHING. Marketing hype and vaporware. That's the only currency you have in this debate.

Those of you continuing to try to debate this just don't get it. The war of words is over. Toshiba and Sony made promises and claims. One of them has stepped up and put their money where there mouth is. We can now make an informed decision as to the veracity of Toshiba and Microsoft's claims about the HD-DVD format and the VC-1 codec. Unless and until Sony does the same thing, the debate is over for the time being. One cannot compare assumptions to fact. Come back when you have the facts.

Lastly, I am not denigrating Sony or Blu-ray. I am only pointing out that Blu-Ray is as yet an unproven technology as is Sony's planned implementation. What I am losing patience with is this incessent yammering proclaiming the virtues of something that does not yet exist in order to divert attention from something that does exist and did meet its stated goals.

AVS has been a forum where technical issues could be discussed by experts in the field as well as informed hobbyists. Discussions have usually been based on sound principles. However, the level of this debate, while always contentious and devisive, has sunk to levels that I find deplorable, even for an internet forum. Anyone can make any wild-ass claim and be taken seriously by others as long as that claim fits their view of the world. When reality finally slaps some of you in the face, I only hope I have the stomach to hang around long enough to see it.

The suggestion that Sony may be deliberately holding back would seem foolish in the light of a format war.

It also contradicts what appears to be the position Sony has taken against Toshiba. Clearly HD DVD is trying to sell to more average audiences (Best Buy, Wall Mart), while the BD pricing looks like it is aimed at the high end (boutiques, high-end early adopters, critical reviewers). If they are trying for these audiences with less than wonderful transfers, Sony is in for a big shock indeed.

I will be very surprised if the BD product is less wonderful than the HD DVD product.

The only MPEG-2 HD material that compares favorably to VC-1 on HD-DVD is D-Theater, and that is at significantly higher bit rate that 18 mb/s, and even then is only comparable or below in quality.

"I, Robot", which is considered as reference material, is encoded at 21 Mb/s (video).

No DVNR was used on those titles. Please go and see for yourself. You don't have to take word of a BD insider when the real thing is here ;). Or, read the reviews in the HD DVD Player section and see the ton of kudos on picture quality. Not one matches what Rio has said. Not even one!
Amir
So you don't see the artifact he described involving that wall?
It does not really matter to me if the artifact is already on the DI/D5 or added during compression. If it's on the disc I will think twice before I buy. And I know for a fact that several DIs are full of DNR issues and the HD discs will show the artifacts in glorious detail. I also expect to see EE on many BR titles from Sony HD masters. It's on the masters.

"I, Robot", which is considered as reference material, is encoded at 21 Mb/s (video).
I have no idea whether this is correct (seems low, but I do not have a way to tell bitrate on my D-VHS deck that I know of). Given that, and given that it is very hard to cross-compare titles, of the four HD-DVDs I own (Serenity, A13, POTO, TLS), I, Robot would only beat out TLS for PQ if I had to pick them. Audio quality is no contest, as one would expect.

What is interesting to me is that I can clearly see limitations of the HD-A1 player with some scenes in Serenity (also for A13). If we can get the clipping issue fixed with the HD-A1, my perception of the PQ for the HD-DVD titles will most likely go up.

The net result: if I, Robot is the quality standard for Sony's MPEG-2 releases (i.e., the best of DTheater), then many, many people will be happy, including Amir and Ben W.

Later,
Bill

I've looked at the D-Theater of I, Robot as well as several other titles on that format. I, Robot has been considered a "reference" because it was subjectively the best many had seen. I no longer consider this title a reference for anything beyond the D-Theater format. One need only compare the MPAA screen with very small text from a D-Theater tape to the same screen at the beginning of any HD-DVD disc. What I see on my system is very obviously greater fine detail on the HD-DVD discs. Whether that is a result of the DVHS deck I am using is, perhaps, part of the reason, but this is something anyone with both D-Theater and HD-DVD can compare.

FYI, the new Windows Media Encoder Studio Edition Beta that'll be available for download sometime today will include the advanced scaling and dithering modes we provide to the studios, so you can give them a shot yourself.

Maybe I'm blind, but as of Monday 11am ET, I'm not seeing a link to download WME Studio Edition.... :confused:

http://www.microsoft.com/windows/windowsmedia/forpros/encoderse/default.aspx

http://www.microsoft.com/windows/windowsmedia/forpros/encoder/default.mspx

"I, Robot", which is considered as reference material, is encoded at 21 Mb/s (video).
Since this topic keeps coming up, here is a bit of food for thought:

* I assume no one here has seen the original master for I Robot. Correct? Then for all you know, some quality was left on the table and you will only find out when better codecs are used for it on these new formats.

* What if it turned out that to beat the I Robot quality on D VHS, VC-1 only required 12-14 Mbit/sec? Remember that current HD DVD titles are rather grainy, making them hard to compress. Some of the future titles look great even at 10 Mbit/sec in VC-1 (although will be encoded at higher rates simply because the capacity is there). Clean content encodes a lot better than others and I Robot is certainly clean.

* You can always find one clip that looks great with some encoding. That doesn't make it a rule. At NAB 4 years ago, we encoded our first HD clip at 720p at just 1.5 Mbit/sec. No, this is not a typo! It really was 1.5 Mbit/sec (we couldn't decode faster than this due to a bug). What was the content? Scenery with slow movement. Should we go around and say that VC-1 looks good at just 1.5 Mbit/sec? Of course not. So unless you have a lot more than I Robot, it is silly to keep pounding on this topic.

* Joe Kane's tests showed that D-VHS is NOT transparent to the source by any stretch. In addition to compression artifacts, it also can not preserve the full frequency response of the source. This was shown clearly on test signals. So no matter how good you think I Robot is, I assure you it is not the same as the source based on objective tests.

As has been said, it is pointless to keep arguing against a ghost here. HD DVD has delivered picture quality many people love. MPEG-2 can only hope to match this performance: http://www.projectorcentral.com/hd-dvd.htm

“However, when the first images from the HD-A1 began to light up the screen Tuesday afternoon, all of my doubts melted away in short order. The image quality was superior to any of the previous demos I'd seen—pure, rock solid, pristine, razor sharp, highly detailed, and virtually artifact-free are just some of the superlatives that come to mind. It actually surpasses broadcast HDTV, for it is in the same class in terms of image resolution, but it is free of the noise and compression artifacts that are part of the broadcast signal. We have used several 720p resolution projectors for our initial look at HD-DVD and the results are beyond any expectation I had. Our associate Bill read my mind when he said "After seeing this it will be hard to look at standard DVD again."

And then this: http://www.thedigitalbits.com/reviewshd/hidefreviews001.html

“Video-wise, the new HD-DVD version of Serenity offers a truly dramatic improvement in quality over the previous DVD's anamorphic widescreen transfer. The improvement in clarity and resolution offered by the new HD image is impressive. The DVD looks terribly soft by comparison - still film-like, certainly, but lacking the sheer level of detail and fidelity of the HD image. This is obvious in many aspects of the picture, for example the background stars when the Serenity is travelling through space. They're generally just a blur if you see them at all on the DVD, but in high-definition, you can count each and every one. Colors are tighter and a hair more accurate, and they're significantly more vibrant. Contrast is a bit better as well, with the darkest and lightest areas of the image retaining more detail in high-definition than on the standard DVD. The two biggest improvements in the HD video (aside from just sheer resolution), are the complete absence of ghosting or edge-enhancement of the kind that so plagues a standard DVD image, and also the almost total lack of visible digital compression artifacting. It's hard to know just how good this high-def video is, and how much better it might get in the future, but Serenity on HD-DVD is a near, if not quite completely, breathtaking experience…. Phantom is a lavish production, with a lush palate and subtle, evocative shadings of light and color. Every little bit of that detail and nuance is reproduced with near perfection in this 1080 presentation. The detail is exquisite, even in the darkest areas of the image, and the shadings are smooth and delicate. The visual image that results is compelling and extraordinarily film-like. Very pleasing indeed.”

You see what I am saying? :)

Amir

Exactly. However, 16Mbps (Serenity) isn't significantly superior to 18Mbps (which apparently Sony plans), does it? VC-1 camp has been saying that the bit-rates that Sony plans to use will not give transparency. Now, from this thread, it looks like 20-22Mbps MPEG2 CBR is giving close to transparency (with some minor artifacts). My extrapolation is that 18Mbps VBR may make it transparent. Do you think it will not? It will also be interesting to see how 16Mbps MPEG2 VBR looks (just out of curiosity).
Sony may plan on using 18 Mbps VBR, but there's no way it's going to look as good as our 16 Mbps VBR on Serenity. Much of the flap about Sony's use of MPEG-2 is that the combination of SL, MPEG-2, and PCM means that Sony will have to use video data rates lower than what experts, myself included, are needed to produce transparent compression.

I want to see advanced codecs on Sony's titles as much as the next AVSer, but I will no longer crucify them because of their choice of MPEG2 for now. Of course in less than two months, we will know for sure.
Yes, the proof will be in the discs. My expectation is that the launch Sony titles will be remarkably less impressive than the launch Warner and Universal titles for HD DVD have been.

Maybe I'm blind, but as of Monday 11am ET, I'm not seeing a link to download WME Studio Edition.... :confused:

No, it just got unexpectedly held up in the final release process somewhere. There's something called "Vista" that seems to have inexplicably been given higher priority than my compression tool beta!

:).

I'm now told sometime this week. I'll post as soon as it's up.

Yes, the proof will be in the discs. My expectation is that the launch Sony titles will be remarkably less impressive than the launch Warner and Universal titles for HD DVD have been.

So far, only Serenity reaches top level video. That's certainly enough to say that VC1 is a superb codec and that combined with HD-DVD, it will reach our expectations.

But it's not better than the best MPEG2 transfers I saw, just as good with lower bitrate (16 VBR against 22 CBR).

That's not idle speculation, just a (subjective, I agree) constatation.

Will Sony MPEG2 transfer look good? There is no way to tell until they are there.

What about what robena has observed? Who is robena and what HAS he observed? Some Japanese broadcast that no one else can verify? Is that the best you can base your conclusions on?

There is a lot said by a lot of people that can not be verified. If what robena has reported turns out to be false, then I will be more cautious trusting him next time! For now, I will believe him.

Assumptions based on ignorance supported by irrational bias. Again, I am at a loss.
Have some coffee. :)


I couldn't care less what the original poster observed. Until there is evidence that others, including me, can observe and verify, there is nothing. That is what some of you in this thread are basing your assumptions and opinions on, NOTHING. Marketing hype and vaporware. That's the only currency you have in this debate.
So, if you start saying what robena has said, I am supposed to believe you because then there will be two guys saying the same thing? How many people should say the same thing before it can be accepted? :rolleyes:

We all have opinions. And then there are facts. When facts come out, we can make a judgment. Until then, I try to follow my instinct when it comes to believing people. By the time Sony's titles come out, we will probably have even better encoded HD DVD titles out. I am just not writing MPEG2 off.

Not everyone on this forum has an agenda. If you cool down, you will get it. This is probably the last post from me in reply to yours. Not that I am angry; I try to avoid people that get angry for no reason. :)

EDIT: For some reason when I quoted the post by ... it became "noname". Is it the tool or some involuntary reflex typing? I don't know. See? Even now, best that came out is "..." :confused: :p

There is a lot said by a lot of people that can not be verified. If what robena has reported turns out to be false, then I will be more cautious trusting him next time! For now, I will believe him.

The word "false" implies a lie and I am not calling anyone a liar. What robena has posted is a subjective opinion of something only he has access to. There is a difference between a lie and heresay. However, I wouldn't take either as fact.

So, if you start saying what robena has said, I am supposed to believe you because then there will be two guys saying the same thing? How many people should say the same thing before it can be accepted?

A lie repeated a thousand times is still a lie. Again, as stated above, not calling anyone a liar. That's a quote I recall from somewhere that serves to make a point.

You can choose to respond or not. Makes no difference. However, I am not angry. I don't get angry at something as trivial as this. I am, however, frustrated that there are so many people so willing to accept any claim without any sort of verifiable evidence just because they want to believe it. That's called religion. I would consider the topic of this forum to fall under the category of science.

* Joe Kane's tests showed that D-VHS is NOT transparent to the source by any stretch. In addition to compression artifacts, it also can not preserve the full frequency response of the source. This was shown clearly on test signals. So no matter how good you think I Robot is, I assure you it is not the same as the source based on objective tests.
Amir
The problems with the frequency response test signals on 1080i D-VHS DVE were due to authoring problems. Just because Joe Kane was unable to QA his own test tape, doesn't mean that D-VHS and MPEG-2 are not capable of delivering full frequency response. In fact, you can download an HD MPEG-2 test pattern from my website with a full amplitude burst at 35 MHz.

http://www.w6rz.net/burst1920.zip

Ron

The problems with the frequency response test signals on 1080i D-VHS DVE were due to authoring problems. Just because Joe Kane was unable to QA his own test tape, doesn't mean that D-VHS and MPEG-2 are not capable of delivering full frequency response. In fact, you can download an HD MPEG-2 test pattern from my website with a full amplitude burst at 35 MHz.

http://www.w6rz.net/burst1920.zip

Ron
The question here was not MPEG-2. The question was D-VHS being as good as it gets (i.e. I Robot). Joe used the same encoders that studios used to make D-VHS content. If something is wrong with them, then all the content in D-VHS suffer the same.

Unless you can tell me that I Robot was encoded with the same tool as your test pattern, there is no evidence that I Robot was transparent to the source.

Amir

You can choose to respond or not. Makes no difference. However, I am not angry. I don't get angry at something as trivial as this.
Good! Then, we are back to debate! Not that it matters! :)

I am, however, frustrated that there are so many people so willing to accept any claim without any sort of verifiable evidence just because they want to believe it.
What this thread has done is to make me open-minded about MPEG2. I still have reserved my judgement until the movies come out. It is the same as you wanting to believe that MPEG2 will always be bad! At this point both your opinion and mine are conjectures.

By the way, the stuff robena observed should be accessible to others too. May be others will chip in with their own observations. There are many AVSers from Japan.

I am going to buy a Blu-ray player, so probably I am biased. I will buy at least one Sony title (more likely 4) initially, and then if I don't like the quality, I will rent other titles! Hopefully, VC-1 encoded titles will follow (from Warner at least). Even though you can not compare different movies, some things will always be common (for instance, macro blocking or lack thereof). If Sony's movies look bad, market will force them to use VC-1 or H.264 AVC. I think I have made a safe bet! :) But I digress.

Unless you can tell me that I Robot was encoded with the same tool as your test pattern, there is no evidence that I Robot was transparent to the source.

Amir
The test patterns on my website were encoded with the LSI Logic HD MPEG-2 real-time encoder (that was co-developed with JVC).

http://www.lsilogic.com/products/video_production__professional_/hdtvxpress.html

http://www.lsil.com/products/customer_success/jvc_hdtv.html

Although it's not the exact same JVC encoder used for all D-Theatre DVHS tapes, it is a direct descendant.

The authoring mistake made on DVE test patterns is a complete canard with regards to the performance of the JVC D-Theater encoder. It was merely the result of the encoder operator cranking in a pre-filter.

Ron

Ron, my assertion still remains. We don't know what knobs where used for I Robot. Maybe they used a pre-filter too. Maybe they did other things. The bottom line is that people hold up I Robot as a reference D-VHS title. All I am saying is that we don't know how good it could have been. A problem we will have with many BD titles if they don't have an alternative using a different codec.

Amir

It is the same as you wanting to believe that MPEG2 will always be bad! At this point both your opinion and mine are conjectures.

I never said I believed MPEG-2 will always be bad. I do, however, believe that based on current state of the art of MPEG-2 encoding that is readily avaiable for evaluation, at the data rates Sony will be required to use for BD-25, MPEG-2 will be inferior to VC-1 encoded discs.

This inferiority could take one of two forms. The first is that the bit rate will simply not be high enough to sustain transparency and will require high frequency filtering to avoid overt compression artifacts, much as has been done by Sony in the past on regular DVD when they wanted to put out a single disc special edition with a large amouint of supplemental content. The second could be that to enable a high enough data rate to approach transparency, discs will be devoid of any supplemental material with only lossy audio tracks available. Either of these scenarios would be inferior to current VC-1 encoded HD-DVD software.

Just to throw one little tidbit out there, I happen to know of at least one studio that plans to release only "bare bones" content on Blu-Ray initially. No supplements at all. File that under "rumor" if you wish as I cannot be more specific.

I am going to buy a Blu-ray player, so probably I am biased.

So, I'm buying a Blu-Ray player also. I'm just not so optimistic that initial Blu-Ray software will be that great. I'm hoping, but the numbers are not adding up.

Guess what, one more little fact to file away regarding my position on this. I'm a Sony dealer. In fact, none of the manufacturers I am a dealer for will have a HD-DVD player out any time soon. I have a financial interest in Blu-Ray being a success. I'm also a practical person. I'm not going to believe something without some firm basis to support it. Right now, I'm not overly optimistic about Blu-Ray, even though it is my best interest that the hype is true.

I guess we have reached the end of our little debate. I am optimistic about the initial MPEG2 titles and you are not! :)

Let's leave it at that for the moment. We can always revisit this once movies are out.

Good post, by the way.

"I, Robot", which is considered as reference material, is encoded at 21 Mb/s (video).


don't forget D-VHS is CBR while BD is VBR

A problem we will have with many BD titles if they don't have an alternative using a different codec.

Amir

It may not be such a big problem afterall :)

It may not be such a big problem afterall :)
I knew I was walking right into that one :D.

Amir

I knew I was walking right into that one :D.

Amir

:) woah, it's like a minefield out there.

A lie repeated a thousand times is still a lie. Again, as stated above, not calling anyone a liar. That's a quote I recall from somewhere that serves to make a point.

You can choose to respond or not. Makes no difference. However, I am not angry. I don't get angry at something as trivial as this. I am, however, frustrated that there are so many people so willing to accept any claim without any sort of verifiable evidence just because they want to believe it. That's called religion. I would consider the topic of this forum to fall under the category of science.

The quote I like is "When the legend becomes fact, print the legend", The Man Who Shot Liberty Valance.

Amir,

Any chance microsoft upload a short demo clip taken from actual HD-DVD content on their media showcase website?

Regarding softness, was it ever confirmed that the Toshibas aren't pre-filtering the interlaced outputs like the HD-DVD specs seemed to indicate as a requirement? Now that people have players, are there any user selectable settings for enabling or disabling that feature?

In order to vertically filter properly to target an interlaced TV you must filter 1080p, not 1080i. To do this in the player you would thus have to inverse telecine the decoded source to 1080p, filter, and then redo the 3:2 pulldown to deliver the 1080i the current Tosh players output.

There is no indication the Tosh HD DVD players have any such sort of processing, as evidenced by their less than stellar 720p output which would also first need 1080p to do correctly. All this is additionally suggested by the lack of any filtering options in the player controls.

The bottom line of all this is that, for HD DVD, any needed vertical filtering for interlace will have to be done at mastering time and, if so, cannot be undone by the player to target a 1080p display.

Hopefully, however, as 1080p displays eventually become more common than 1080i displays the studios will feel less need to filter at all.

- Tom

Can anyone speculate whether the video on an MPEG-2 encoded 25GB BD discs should look better than what's on HBO-HD at least? The video on HBO-HD is MPEG-2, correct?

Can anyone speculate whether the video on an MPEG-2 encoded 25GB BD discs should look better than what's on HBO-HD at least? The video on HBO-HD is MPEG-2, correct?

It should look a good deal better than whats on HBO-HD.

In order to vertically filter properly to target an interlaced TV you must filter 1080p, not 1080i. To do this in the player you would thus have to inverse telecine the decoded source to 1080p, filter, and then redo the 3:2 pulldown to deliver the 1080i the current Tosh players output.

There is no indication the Tosh HD DVD players have any such sort of processing, as evidenced by their less than stellar 720p output which would also first need 1080p to do correctly. All this is additionally suggested by the lack of any filtering options in the player controls.

The bottom line of all this is that, for HD DVD, any needed vertical filtering for interlace will have to be done at mastering time and, if so, cannot be undone by the player to target a 1080p display.

Hopefully, however, as 1080p displays eventually become more common than 1080i displays the studios will feel less need to filter at all.

- Tom

Since HD-DVD movies are stored as 1080p24; why would inverse telecine be needed in the player? The original 1080p source could simply be filtered before the pulldown is applied. If the player and all 1080p displays supported 1080p connections so 1080i was only used for CRT displays then it would make sense to filter the 1080i output to avoid interlace artifacts on the CRT (and also to help discourage the studios from filtering at the source). So now I wonder if a current progressive display using inverse telecine of the 1080i connection would have the vertical frequency response reduced by filtering in the player.

So now I wonder if a current progressive display using inverse telecine of the 1080i connection would have the vertical frequency response reduced by filtering in the player.

So far, with the 5 discs I have watched, I did not see what I expected to be a significant increase of definition compared to filtered MPEG2.

Serenity was gorgeous, but not more detailed than the best I have already seen.

Ron, my assertion still remains. We don't know what knobs where used for I Robot. Maybe they used a pre-filter too. Maybe they did other things. The bottom line is that people hold up I Robot as a reference D-VHS title. All I am saying is that we don't know how good it could have been. A problem we will have with many BD titles if they don't have an alternative using a different codec.

Amir

Better get cracking with Sonic then and get your "Pro" VC-1 encoder intergrated into their BD authoring suite ASAP. Waiting for Sony to roll out the red carpet for VC-1 isn't in the tea leaves... ;)

b2b

Any chance microsoft upload a short demo clip taken from actual HD-DVD content on their media showcase website?
Licensing for the actual HD-DVD content is a challenge, and of course we don't have a HD DVD player yet. Now that NAB is over I'm trying to fit in some time to make some samples muxed into a WMV file that you can play back, with VC-1 video encoded to the HD DVD spec.

Better get cracking with Sonic then and get your "Pro" VC-1 encoder intergrated into their BD authoring suite ASAP. Waiting for Sony to roll out the red carpet for VC-1 isn't in the tea leaves... ;)

We were actualy demoing Sonic's VC-1 encoder in our both at NAB, FYI.

We were actualy demoing Sonic's VC-1 encoder in our both at NAB, FYI.
Well, I was going by what Amir said in this post
Quote:
Originally Posted by admonish
Who has access to the 'Pro VC-1 Encoder'?

Currently we are offering it to studios/post production houses that have the ability to use it for HD DVD/BD production. The tool requires training and hand holding initially so it is not ready for general availability. Same is true btw to a larger extent in the case of the authoring tools themselves.

Quote:
Does Scenarist clearly state that they have the 'J6P VC-1 Encoder'?

I don't know what it has in it. All I can tell you what it does NOT have in it (i.e. our pro encoder). For all I know, they could have written their own encoder from VC-1 spec, licensed it from someone else, or use the standard one in Windows. In the latter case at least, it is better than what you are implying. Our Pro encoder however, in addition to having better picture quality, also supports advanced features such as ability to use up to 8 CPUs to do parallel encoding and other workflow related features. These things are just as important as better picture quality.

Amir

My impression is that the only people using the "Pro" version of the encoder are the post houses doing work for HD-DVD title production. Am I wrong to assume getting Sonic's product integrated with the production quality "Pro" VC-1 encoder isn't going to happen any time soon? Or is it still a moving target with tweaking still going on with the encoder?

b2b

Licensing for the actual HD-DVD content is a challenge, and of course we don't have a HD DVD player yet. Now that NAB is over I'm trying to fit in some time to make some samples muxed into a WMV file that you can play back, with VC-1 video encoded to the HD DVD spec.
Screw clips! We need a test disc to calibrate our displays, and ASAP! (Stacey said he and Don were thinking of one, and this should be something that gets fast-tracked from a Demo point of view...).

oh, yeah, please?

Later,
Bill

My impression is that the only people using the "Pro" version of the encoder are the post houses doing work for HD-DVD title production. Am I wrong to assume getting Sonic's product integrated with the production quality "Pro" VC-1 encoder isn't going to happen any time soon? Or is it still a moving target with tweaking still going on with the encoder?

The Sonic encoder is a commerical product, seperate from the technology we've provided to the studios for the launch titles. We helped them with their codec implementation, of course. :)

I'll let Sonic speak for themselves, but they were announcing immediate availabilty of both their compression and HD DVD authoring systems at NAB. Nice UI, I thought.

So far, with the 5 discs I have watched, I did not see what I expected to be a significant increase of definition compared to filtered MPEG2.

Serenity was gorgeous, but not more detailed than the best I have already seen.

BD Troll!

BD Troll!You might want to review the forum posting rules. The mods have a very low tolerence level for name calling..
Forum Rules: Read This Before You Post
Welcome to the AVS HDTV Forums:

special rules apply to posting here: please observe them:

-No bashing other members
-Please no fighting or trolling
-if you disagree with something posted: challenge the information; not the poster
-make sure your information is accurate; cite the facts on which you base your post: if it is your opinion, qualify it as such
-are you an industry insider? identify yourself as such and do not use AVS to bash your competion

AVS is a technical forum: we expect posters to maintain a high techical and ethical standard in your posts

AVS has Moderators who volunteer to enforce these rules: Moderators can edit or delete your post and even recommend the suspension of posters who continually violate AVS rules

If you see a post that violates rules or is spam, contains bashing, trolling, etc: please do not respond to it: use the 'report this post to a Moderator' button and let the mods handle it

b2b

You might want to review the forum posting rules. The mods have a very low tolerence level for name calling..

b2b


I get called a troll posted-as-a-reply a lot recently, whenever I bring up a negative about HD-DVD or BD. Sorry, forgot my smiley which makes everything ok, i guess.

:)

From Wowow, Harry Potter 3 and Black Hawk Down (completely stunning in some low light scenes, not the grainy ones).

From BS/Hi, The Wrath of Khan.

From HBO, Drop Zone despite an average bitrate. And The Naked Gun 33 1/3.

I think that the initial photography and transfer are going to be the limiting factor, not the codecs.
Where the hell are you getting Wowow and BShi HD J-stations?? Are you living in Japan??? If you're living in the states, how did you get it. Could you provide more info? I'm dying to get one of those. I know what those HD stations can do... they provide the best HD quality you can get...

Please please reply. Thank you........

Bill Hunt a couple of weeks into the HD DVD life ... Not too happy...
What this does have to do with VC-1 vs MPEG-2? If anything, he does say the picture quality is just fine with VC-1. Since both VC-1 and MPEG-2 are available in either format, it seems silly to bring up stuff like this anyway. To say nothing of the fact that that there are entire threads on that review already and we have the long discussion thread...

If we have exhusted the available data on this topic, then we should let it die. Let's not find ways to create more parallel arguments :).

Amir

What this does have to do with VC-1 vs MPEG-2? If anything, he does say the picture quality is just fine with VC-1. Since both VC-1 and MPEG-2 are available in either format, it seems silly to bring up stuff like this anyway. To say nothing of the fact that that there are entire threads on that review already and we have the long discussion thread...

If we have exhusted the available data on this topic, then we should let it die. Let's not find ways to create more parallel arguments :).

Amir

Well I wasn't going to bring it up....but... ;) Why the less than enthusiastic attitude about getting the "Pro" encoder working with Sonic's Pro authoring suite for BD? While it might not be the best solution for tier one studios, it seems to be the right product for lower profile productions especially for getting the very best quality VC-1 content out on BD. You wouldn't want VC-1 looking less than its very best irregardless of platform would you?

b2b

Me would like an answer please.... Post #131

Me would like an answer please.... Post #131

I sent you a PM.

What this does have to do with VC-1 vs MPEG-2? If anything, he does say the picture quality is just fine with VC-1. Since both VC-1 and MPEG-2 are available in either format, it seems silly to bring up stuff like this anyway. To say nothing of the fact that that there are entire threads on that review already and we have the long discussion thread...

If we have exhusted the available data on this topic, then we should let it die. Let's not find ways to create more parallel arguments :).

Amir

Sheesh... Sorry, posted it in the wrong thread. Was meant for Monster Thread III.

Why the less than enthusiastic attitude about getting the "Pro" encoder working with Sonic's Pro authoring suite for BD?
Why do you think we are "less than enthused" about partnership with them? While I can not disclose confidential information regarding partner activities in general, I have been very clear that we are working with them.

While it might not be the best solution for tier one studios, it seems to be the right product for lower profile productions especially for getting the very best quality VC-1 content out on BD. You wouldn't want VC-1 looking less than its very best irregardless of platform would you?

b2b
I really don’t understand this line of questioning from BD folks. Isn’t there really two options here?

1. That I am telling the truth that we are trying hard to get VC-1 encoding done in BD format.

2. That we don’t really want to support VC-1 in BD.

If you believe in #1, then you already have your answer. If you believe in #2, why on earth do you think asking me about this topic and/or making suggestions does any good?

Amir

Hey AMIRM,

This is off topic but since you are a MC12 owner. Lexicon announced their new MC12 HD unit.

Just thought I would distract you folks from this thrilling thread for a moment

Hey AMIRM,

This is off topic but since you are a MC12 owner. Lexicon announced their new MC12 HD unit.

Just thought I would distract you folks from this thrilling thread for a moment
Cool! Just looked it up. Alas, it doesn't say anything about an upgrade. I guess it is time to go and beg Madrigal execs for a good price on yet another processor! :).

Amir

Amir,

I think that most of us BD folks go with number one.

Amir,

I think that most of us BD folks go with number one.

LOL, even though that is probably not true

Amir,

I think that most of us BD folks go with number one.
Much appreciated David. Hope to disappoint those who do not one day :).

Amir

Much appreciated David. Hope to disappoint those who do not one day :).
I will be very happy if that day happens to be May 23rd! :)

We could then close this thread!

Why do you think we are "less than enthused" about partnership with them? While I can not disclose confidential information regarding partner activities in general, I have been very clear that we are working with them.


I really don’t understand this line of questioning from BD folks. Isn’t there really two options here?

1. That I am telling the truth that we are trying hard to get VC-1 encoding done in BD format.

2. That we don’t really want to support VC-1 in BD.

If you believe in #1, then you already have your answer. If you believe in #2, why on earth do you think asking me about this topic and/or making suggestions does any good?

Amir
Maybe there's a third option. Seeing Pro quality VC-1 in third party BD and HD-DVD authoring tools would be the best option of all..

Oh no, what am I saying, I'm a "BD folk". :eek:

Make that a Pro quality H.264 encoder in BD and HD-DVD authoring tools... :)

b2b

Maybe there's a third option. Seeing Pro quality VC-1 in third party BD and HD-DVD authoring tools would be the best option of all..

Oh no, what am I saying, I'm a "BD folk". :eek:

Make that a Pro quality H.264 encoder in BD and HD-DVD authoring tools... :)

Without Film Grain Replacement in BD, it's an open question of whether there could be a pro quality H.264 encoder for film source content...

But yes, we certainly help vendors of HD optical authoring tools across the price spectrum integrate professional grade VC-1 encoding with their products.

ben,

are you stating that the film-grain challenges different codecs in different ways (may be a problem for one but not the other)?

Please ellaborate...

thanks,

dave

Without Film Grain Replacement in BD, it's an open question of whether there could be a pro quality H.264 encoder for film source content...



Ben -

Sorry, you lost me on that one. Are you referring to the claim AVC does not do well on film without filtering? I'm not sure we all accept that.

BTW, I'm a fan of (moderate) FGT in theory though I've never actually had a chance to see the results in AVC or (of course) BD.

- Tom

Hopefully they just get rid of the grain eventually except when it's there for intended artistic reasons.

Transparency is good when the source is great, but I want something better, perhaps hand edited frame by frame by hand, if it's better than "transparent".

Without Film Grain Replacement in BD, it's an open question of whether there could be a pro quality H.264 encoder for film source content...

But yes, we certainly help vendors of HD optical authoring tools across the price spectrum integrate professional grade VC-1 encoding with their products.

Well, in search for examples of your "doubts" I ran across these which are captures of the H.264 Serenity trailers.

These two are of particular interest because of striking quality compared to the reported images taken with a digital camera of the actual Serenity HD-DVD VC-1 encoded disc in this post (http://www.avsforum.com/avs-vb/showthread.php?p=7536195&&#post7536195)
The next two images are reported to be taken with a digital camera from the VC-1 HD-DVD disc.

http://audioworx.virtualave.net/Serenity5.jpg
and
http://audioworx.virtualave.net/Serenity6.jpg

The next two are the H.264 captures from the Apple trailer..

http://homepage.mac.com/jharrell/serenity/serenity-3.jpg
and
http://homepage.mac.com/jharrell/serenity/serenity-4.jpg

hum... looks pretty close to me, maybe too close?

Here is the link where I found the H.264 captures. It was dated April 29, 2005, over a year ago..
http://theshapeofdays.com/h264/

b2b

B2b good catch.

B2b good catch.

Well it wouldn't have been really interesting to me except that those images that were reportedly made with a digital camera were what convinced me that what I saw with the demo setup at Magnolia was due to a poor display and/or improper calibration. Now I have my doubts again with this seemingly bogus attempt to portray HD-DVD quality with doctored "screen shots".

And I'm NOT laying any of this at Ben's or Amir's doorstep. The questions need to answered by the person that made the claims...

b2b

Those shots are what are making me get a HD-DVd player and serenity even though I didn't think the movie was great.

I hope that VC-1 can come close to waht the trailer looks like.

Maybee Amir or Ben can show actual screengrabs from the movie to show us comparisons.

Sorry, you lost me on that one. Are you referring to the claim AVC does not do well on film without filtering? I'm not sure we all accept that.

BTW, I'm a fan of (moderate) FGT in theory though I've never actually had a chance to see the results in AVC or (of course) BD.

My point is that AVC needs film grain synthesis at VC-1 bitrates to reproduce film grain well.

My point is that AVC needs film grain synthesis at VC-1 bitrates to reproduce film grain well.

Well there is this screen capture from the Serenity H.264 trailer that seems to be doing film grain pretty well.

http://homepage.mac.com/jharrell/serenity/serenity-5.jpg

b2b

Hopefully they just get rid of the grain eventually except when it's there for intended artistic reasons.

Transparency is good when the source is great, but I want something better, perhaps hand edited frame by frame by hand, if it's better than "transparent".
I think I agree with you :). The goal is the most accurate representation of the creators' intent. If they wanted a particular grain, that should be in there. If they didn't, it shouldn't. There's certainly masters for older film where grain and other artifacts have come in through being reprinted and such, and cleaning that up is definitley a good thing. Of course, in a lot of cases for older movies, the creators are dead so we have to guess how they would have wanted it to look.

Anyway, that's really an upstream question for the film restoration guys. On the codec level, we need to be able to handle whatever they throw at us.

I'm curious to see if the film aesthetic evolves away from grain, like digital photography does. Of course, grain makes compositing and other VFX quite a bit easier to blend in.

hum... looks pretty close to me, maybe too close?I think somebody mentioned that the first one is swapped left-to-right from what is actually in the movie also.

--Darin

Well there is this screen capture from the Serenity H.264 trailer that seems to be doing film grain pretty well.

It's not horrible by any means, but even on my 17" laptop screen I can certainly see quantization errors in the grain. And those kinds of errors typically look worse in motion than on a still frame, since they tend to change frame-by-frame.

I'd never argue that H.264 can't produce a pleasing image -my argument is that transparency with film grain content is a challenge for the codec, hence the use of film-grain replacment in HD DVD (but not BD).

I can also say that the complexity of using film grain replacement definitely has made VC-1 a more compelling option.

It's not horrible by any means, but even on my 17" laptop screen I can certainly see quantization errors in the grain. And those kinds of errors typically look worse in motion than on a still frame, since they tend to change frame-by-frame.

I'd never argue that H.264 can't produce a pleasing image -my argument is that transparency with film grain content is a challenge for the codec, hence the use of film-grain replacment in HD DVD (but not BD).

I can also say that the complexity of using film grain replacement definitely has made VC-1 a more compelling option.

I'm all confused now. My understanding is that VC-1 has no provisions for FGT according to the spec. But H.264 does and it is up to the producer of the title to use it or not. So correct me if I'm wrong, that VC-1 has no option to use FGT. But H.264 does and it is optional depending on the wishes of the content producer, at least for BD?

b2b

My point is that AVC needs film grain synthesis at VC-1 bitrates to reproduce film grain well.Ben, just curious have you ever seen the use of FGT with MPEG-4 AVC video before? I have heard forum members reactions to the demonstrations of it at various shows and to say the least "bad looking" is the best thing I heard. Also MPEG-4 AVC can reproduce film grain so let us keep the VC-1 promotion reasonable in this thread.


I'm all confused now. My understanding is that VC-1 has no provisions for FGT according to the spec. But H.264 does and it is up to the producer of the title to use it or not. So correct me if I'm wrong, that VC-1 has no option to use FGT. But H.264 does and it is optional depending on the wishes of the content producer, at least for BD?FGT was basically the reason that Thomson switched to HD-DVD since the results of it were not that good from what I have heard. The BDA rejected it as an unnecessary addition so Thomson took it to the DVD Forum where it was accepted. Technically Thomson wanted it to be used on both advanced video codecs, but I don't think Microsoft had any interest in doing that so it ended up being added only to MPEG-4 AVC.

I think I agree with you . The goal is the most accurate representation of the creators' intent. If they wanted a particular grain, that should be in there. If they didn't, it shouldn't. There's certainly masters for older film where grain and other artifacts have come in through being reprinted and such, and cleaning that up is definitley a good thing. Of course, in a lot of cases for older movies, the creators are dead so we have to guess how they would have wanted it to look.

Anyway, that's really an upstream question for the film restoration guys. On the codec level, we need to be able to handle whatever they throw at us.

I'm curious to see if the film aesthetic evolves away from grain, like digital photography does. Of course, grain makes compositing and other VFX quite a bit easier to blend in.

Agreed, Agreed, and Agreed!

dave :)

I'm all confused now. My understanding is that VC-1 has no provisions for FGT according to the spec. But H.264 does and it is up to the producer of the title to use it or not. So correct me if I'm wrong, that VC-1 has no option to use FGT. But H.264 does and it is optional depending on the wishes of the content producer, at least for BD?

VC-1 doesn't need FGT, because its basic design does better with film grain.

H.264 has FGT available in HD DVD, but not in Blu-ray.

VC-1 doesn't need FGT, because its basic design does better with film grain.

H.264 has FGT available in HD DVD, but not in Blu-ray.
Correct. I'll back you on both those statements.

VC-1 doesn't need FGT, because its basic design does better with film grain.Neither advanced video codec needed FGT and all the reactions to FGT makes it sound like a rather bad technology. Also I remember that Thomson argued for it's use in both MPEG-4 AVC and VC-1.


H.264 has FGT available in HD DVD, but not in Blu-ray.True, and to further expand on what I said earlier I should point out that Thomson was the company that made FGT and owns the patents to it. Thomson wanted it to be part of both Blu-ray and HD-DVD, but in the end the BDA rejected it since as Keith notes (http://www.avsforum.com/avs-vb/showthread.php?p=5524720&&#post5524720) it was even a pretty bad implementation of film grain model technology. Thomson than took it to the DVD Forum which accepted it and made it part of the HD-DVD standard. Thomson wanted it to be used with both MPEG-4 AVC and VC-1, but ended up only being able to add it to MPEG-4 AVC. The stated goal of FGT is to allow for lower bit rate encoded video by having the film grain "removed" and than randomly adding it back into the video on playback. The only demonstrations ever made of FGT looked horrible and even Stacey agreed (http://www.avsforum.com/avs-vb/showthread.php?p=5816533&&#post5816533) that it is a pretty bad technology. This addition of FGT to HD-DVD more than any other reason is why Thomson supports the format.

The stated goal of FGT is to allow for lower bit rate encoded video by having the film grain "removed" and than randomly adding it back into the video on playback.

If that's the way the algorithm worked, then why are we even taking it seriously in these discussions (thanks for clarifying this excellent point). That's not a videophile feature...that's not about fidelity or transparency at all.

It's not horrible by any means, but even on my 17" laptop screen I can certainly see quantization errors in the grain. And those kinds of errors typically look worse in motion than on a still frame, since they tend to change frame-by-frame.

I'd never argue that H.264 can't produce a pleasing image -my argument is that transparency with film grain content is a challenge for the codec, hence the use of film-grain replacment in HD DVD (but not BD).

I can also say that the complexity of using film grain replacement definitely has made VC-1 a more compelling option.

Considering the capture was from a 8Mbps H.264 I would say it looks exceptionally good (and that's considering the effect of converting it to JPEG format).

b2b

This may be off-topic but I'm curious as well....

Where the hell are you getting Wowow and BShi HD J-stations?? Are you living in Japan??? If you're living in the states, how did you get it. Could you provide more info? I'm dying to get one of those. I know what those HD stations can do... they provide the best HD quality you can get...

Please please reply. Thank you........

If that's the way the algorithm worked, then why are we even taking it seriously in these discussions (thanks for clarifying this excellent point). That's not a videophile feature...that's not about fidelity or transparency at all.
Richard is not giving FGT proper treatment :). His favorite codec committee, H.264/MPEG-4 AVC put the facilities for FGT in that codec. So if he thinks FGT is a bad idea, he should rethink his overall choice of codecs :).

Seriously, FGT provides a trade off between compression artifacts and film grain fidelity. For some content, at some bit rates, FGT produces better picture in AVC. Its film grain btw, is not random as Richard says. At encode time, FGT models (analyzes) the grain present in the original film, and attempts to simulate it with similar pattern and playback time. I am confident someone could come up with content that looks better with FGT than without and vice versa.

As Ben says, with VC-1, we don't need it because we tend to encode grain better than AVC.

I am surprised by the history of events that Richard claims to know regarding this technology. Much of the history of FGT is not public. Specifically, BDA dealings are very private so he would have no idea of the actual timing there. It is however a fair assumption that Thomson was probably not happy that FGT was not approved in BDA, as would Dolby for example if its codec was not selected in DVD Forum.

Amir

Richard is not giving FGT proper treatment. His favorite codec committee, H.264/MPEG-4 AVC put the facilities for FGT in that codec. So if he thinks FGT is a bad idea, he should rethink his overall choice of codecs.I never said I liked everything that was done by the JVT committee, but I would point out that FGT was never actually made part of the video codec. In fact from what I have heard the main reason support for FGT was added to MPEG-4 AVC HP was because of Thomson.


Seriously, FGT provides a trade off between compression artifacts and film grain fidelity.In my opinion taking away film grain and than adding it back to the video on decode somewhat kills any argument for it being about film grain fidelity.


For some content, at some bit rates, FGT produces better picture in AVC.Amir, aren't you being just a tad vague? Also if this is true why have I never heard of any demonstration of FGT that did not look horrible?


Its film grain btw, is not random as Richard says. At encode time, FGT models (analyzes) the grain present in the original film, and attempts to simulate it with similar pattern and playback time."Attempts to" and "simulate" though are two rather important terms to note. Also Amir it was Stacey who said this (http://www.avsforum.com/avs-vb/showthread.php?p=5816533&&#post5816533): "FGT is also frame based, not region based. If, for example, grain was removed from a character in a film for artistic reasons, FGT would add grain where it does not belong."


I am surprised by the history of events that Richard claims to know regarding this technology. Much of the history of FGT is not public.Amir, I remember the discussions that occurred when FGT was posted about in the first "Latest HiDef DVD News" thread. At that time you, Stacey, and Keith talked a good deal about what FGT was and how it worked.


It is however a fair assumption that Thomson was probably not happy that FGT was not approved in BDA, as would Dolby for example if its codec was not selected in DVD Forum.Certainly, I have read (http://www.eetimes.com/showArticle.jhtml?articleID=59301182) that Thomson took years to create FGT and that they even helped to add support for it in MPEG-4 AVC HP to increase the chance that their technology would be used.

I never said I liked everything that was done by the JVT committee, but I would point out that FGT was never actually made part of the video codec. In fact from what I have heard the main reason support for FGT was added to MPEG-4 AVC HP was because of Thomson.
Amazing power Thomson has. First in HD DVD. Then in Joint ITU/MPEG group ;).

In my opinion taking away film grain and than adding it back to the video on decode somewhat kills any argument for it being about film grain fidelity.
Again, this is not about film grain for the sake of film grain. It is about you liking compression artifacts such as macroblocks more than grain that doesn't match its original. Either way though, I assume you have not seen FGT in action so I hope you only go so far in your assertions before you do.

Amir, aren't you being just a tad vague? Also if this is true why have I never heard of any demonstration of FGT that did not look horrible?
Tad vague? Folks generalize codecs all the time. "18 Mbit/sec average rate MPEG-2 should look just fine." Have you seen 18 Mbit/sec MPEG-2? No. Yet you believe that story but want to see more detailed proof for FGT?

I am not a fan of FGT and would be the last person to say something good about it. But we know of tests that show that it improves overall picture quality with AVC. The fact that you haven't heard about such tests has no bearing on the data here since you are not in the business and not everything shows up in the press.

"Attempts to" and "simulate" though are two rather important terms to note.
This is not as difficult a problem as you think. Film stock is film stock. There is not infinite variety here. All the algorithm has to do is identify the type and it is able to simulate it reasonably well at the other end. The damage if anything, will come from filtering it at the source. But with AVC, you may not be able to tell the difference one way or the other :).

Also Amir it was Stacey who said this (http://www.avsforum.com/avs-vb/showthread.php?p=5816533&&#post5816533): "FGT is also frame based, not region based. If, for example, grain was removed from a character in a film for artistic reasons, FGT would add grain where it does not belong."
Yes, FGT can degrade the source. I said this would happen and indeed, does happen. It is just that on noisy content at lower bit rates, it can sometimes improve quality. It is simply another tool at the disposal of an AVC compressionist for HD DVD. If it does some good, it can be used. If it does not, it won't. It is not like one is being forced to use it.

Amir, I remember the discussions that occurred when FGT was posted about in the first "Latest HiDef DVD News" thread. At that time you, Stacey, and Keith talked a good deal about what FGT was and how it worked.
Which doesn't invalidate what I said.

Amir

Amir,

but as long as we have datarates on HD DVD and Blu-ray that can support "literal" encoding of film-grain with VC1, then I'm assuming we all agree that this would be the preferred option.

Any low-bit-rate-optimized algorithm might be good for web casts and other media types where fidelity is not the primary goal and bandwidth is severely limited...and maybe "film grain replacement" approaches would work there since the image is already compromised in many other ways.

But-

Niether of those criteria should apply to BD or HD DVD!

It's not horrible by any means, but even on my 17" laptop screen I can certainly see quantization errors in the grain. And those kinds of errors typically look worse in motion than on a still frame, since they tend to change frame-by-frame.

I'd never argue that H.264 can't produce a pleasing image -my argument is that transparency with film grain content is a challenge for the codec, hence the use of film-grain replacment in HD DVD (but not BD).

I can also say that the complexity of using film grain replacement definitely has made VC-1 a more compelling option.

You're talking about the original trailer?

JPEG itself can introduce such errors, right?

Gary

but as long as we have datarates on HD DVD and Blu-ray that can support "literal" encoding of film-grain with VC1, then I'm assuming we all agree that this would be the preferred option.
We definitely do :). And if we do agree on this, then we also have to agree that we don't need BD-50, peak video rates of 40 Mbit/sec, etc. :) I don't want to debate these points btw (please Richard, don't get started on this). But if we are just going to assign scores to paper advantages on each side, FGT on the HD DVD deserves a mention too.

Any low-bit-rate-optimized algorithm might be good for web casts and other media types where fidelity is not the primary goal and bandwidth is severely limited...
Internet video does not need FGT as one would filter all of that out anyway. FGT is really for say, 10-15 Mbit/sec AVC. And to the extent someone is keen on using AVC, this data range is smack in the middle of where the content owners want to go.

Amir

We definitely do :). And if we do agree on this, then we also have to agree that we don't need BD-50, peak video rates of 40 Mbit/sec, etc. :)

Didn't you say a few days ago that you had stopped arguing on the pros and cons of BD vs HD DVD? :rolleyes:

Or is it only when it fits you? ;)

Did you read the next two sentences under what you quoted? Where he says "I don't want to debate these points" and mentions that he uses them for the limited point of showing how assigning points to those sorts of things is meaningless?

And if we do agree on this, then we also have to agree that we don't need BD-50, peak video rates of 40 Mbit/sec, etc.

It depends. Does that "peak" rate refer to simultaneous video streams such as multi-angle? If so...then the more the better. That way 2 or even 3 simultateous "angles" could all be presented side-by-side in VC1 allowing the user to switch to video commentary or special-effects work prints during movie play.

If we're talking about BD's peak-rate in general, then that also includes bandwidth for audio.

I'd like for content providers to be able to give me DTS HD/DTHD 6.1 feature *and* music-only tracks... and maybe some high quality DD+ tracks in other languages.

The more bandwidth the better.

Amir, wouldn't you agree?

You're talking about the original trailer?

JPEG itself can introduce such errors, right?


I KVM over to my G5, and find the Serenity 1080p on my desktop!

I see similar artifacts in the source there.

And also, those frames aren't indicative of the quality of that trailer overall - there are much more visible artifacts in other regions.

This is copyrighted material, so I personally am not going to post screen shots, but here's the QuickTime time code for a few not-so-good frames I found in a random 10 second region:

14.34
19.89
20.51
22.06

Of course, we certainly don't expect transparency out of a 8 Mbps Main Profile H.264

In the latest issue of S&V Sony says they will be using MPEG-2 with an average data-rate of 18Mbs. That isn't much better than what I am getting now from my cable. Is 18Mbs enough to not have artifacts in fast motion scenes?

In the latest issue of S&V Sony says they will be using MPEG-2 with an average data-rate of 18Mbs. That isn't much better than what I am getting now from my cable. Is 18Mbs enough to not have artifacts in fast motion scenes?

Cable is only 18mbs (on the good stuff), and that's a constant bit rate (CBR). The optical disc formats are variable bit rates (VBR) meaning they can use more bits for the difficult scenes. So that's an average of 18 Mbs.

It should be much better than cable HD.

ATSC is 19.2 for both audio and video, and generally wastes a lot of space with null packets (guesstimates from material I have recorded are ~1/4 of the space). Sony's VBR MPEG-2 should equal or exceed broadcast HD (though that is not saying much for fast motion/pans).

Cable is only 18mbs (on the good stuff), and that's a constant bit rate (CBR). The optical disc formats are variable bit rates (VBR) meaning they can use more bits for the difficult scenes. So that's an average of 18 Mbs.

It should be much better than cable HD.
Agreed.

My concern with 18 Mbps VBR isn't about hard scenes being bad (there's plenty of headroom for peaks, even with PCM audio), but about not having enough total bits for high consistant quality.

How about the encoder itself ? I read here that they are using realtime mpeg2 encoders. Can that eliminate motion artifacts in tough scenes ... ? I hope they don't go for quantity over quality ...

How about the encoder itself ? I read here that they are using realtime mpeg2 encoders. Can that eliminate motion artifacts in tough scenes ... ? I hope they don't go for quantity over quality ...

Looking at the imminent PS3 I would say that Sony is definitely going for quantity over quality.

The PS3 just ain't no ES series product. It is about as cheap as it can get. And Sony does that to sell huge volumes (at least that's what they hope).

So there you have your answer.

:rolleyes:

no, that's not your answer.

Sony makas a variety of products.

Have you seen the Qualia or Ruby 1080P SXRD projector? why not pick that product instead?

:rolleyes:

no, that's not your answer.

Sony makas a variety of products.

Have you seen the Qualia or Ruby 1080P SXRD projector? why not pick that product instead?

You apparently did not even bother to read my comment completely.

Where did I generalize that Sony does not produce high quality products?

My statement was only with regard to the PS3. That will be a cheap plastic box.

I also stated that it is not even close to the quality of Sony's ES series.

That should have told you that I am familiar with Sony products that are built to higher standards than game consoles/a.k.a. toys.

The Qualia range is also quite impressive, but nowhere near the VPH-G90 or VPH-G70 that were truly built like tanks.

I also stated that it is not even close to the quality of Sony's ES series.
I guess you'll have to go the for the PS3 ES Series. Sony has also given permission to many higher end companies like Meridian, Theta, and Classe to build special "high quality" versions of the PS3, but they will be based off of a Pioneer BD player. :D

My statement was only with regard to the PS3. That will be a cheap plastic box.

I also stated that it is not even close to the quality of Sony's ES series.

That should have told you that I am familiar with Sony products that are built to higher standards than game consoles/a.k.a. toys.

The Qualia range is also quite impressive, but nowhere near the VPH-G90 or VPH-G70 that were truly built like tanks.

I find the notion that the "quality" of a high-tech product can be judged by whether the case is plastic or metallic to be very odd. Not only does it have nothing to do with how technologically sophisticated the product is, it also has nothing to do with how reliable it is.

The Source of this Encode was a 19 Mbit Mpeg-2 1920x1080p encode from Ben Waggoner, it was Balanced for Decoding Encoding and Quality Complexity (High Profile) (yes it is possible) useing the Open Source X264 H.264 Encoder.

http://rapidshare.de/files/19769944/The-Island_HD-Trailer-High_Profile-5.1-HE-AAC-8000kbps.mp4.html

keep in mind though this is a "second hand" Source nothing with what some of you work here but it preserves the problems and details of the Source well and that @ 8 Mbit.

I find the notion that the "quality" of a high-tech product can be judged by whether the case is plastic or metallic to be very odd. Not only does it have nothing to do with how technologically sophisticated the product is, it also has nothing to do with how reliable it is.

Indeed. Was there ever released a ES type quality gaming device?? I mean, ever? Why would any company release one? To sell a dozen units? It just doesn't make sense, on any level... You need a sophisticated device, better than whats already on the market and at a price people are willing to pay. (As in cheap enough to sell how ever many units you are able to produce.) Quality needs to be good enough to avoid returns of defective units.

I think the PS3 will deliver. I'm not asking Sony for ES type quality. I don't want to pay more than what I need to...

The Source of this Encode was a 19 Mbit Mpeg-2 1920x1080p encode from Ben Waggoner, it was Balanced for Decoding Encoding and Quality Complexity (High Profile) (yes it is possible) useing the Open Source X264 H.264 Encoder.

http://rapidshare.de/files/19769944/The-Island_HD-Trailer-High_Profile-5.1-HE-AAC-8000kbps.mp4.html

keep in mind though this is a "second hand" Source nothing with what some of you work here but it preserves the problems and details of the Source well and that @ 8 Mbit.
Er, the license I got from Dreamworks on that does NOT allow third parties to reencode and distribute it.

Also, that clip was encoded with a MPEG-2 encoder that does adaptive softening, so it's not a particularly useful test for the D5 workflow.

Looking at the imminent PS3 I would say that Sony is definitely going for quantity over quality.

Well PS3 is not even imminent ;)

Anyway, my question is - will Sony churn out as many movies as possible i.e. prefer quantity over a few well made releases. We can all guess - but we won't really have an answer ...

How about the encoder itself ? I read here that they are using realtime mpeg2 encoders. Can that eliminate motion artifacts in tough scenes ... ? I hope they don't go for quantity over quality ...
I haven't used it myself, but I've heard it's pretty good, but not much better than pretty good. Hardware encoders need to do everything they're going to do with a frame of video in real time, where a software encoder is able to slow down as needed for complex frames.

Hardware encoders definitely had the quality edge over software in the 90's, but I'd say the best software exceeds the best hardware now. I did a bunch of SD MPEG-2 testing last fall, and this was definitely true there.

I haven't used it myself, but I've heard it's pretty good, but not much better than pretty good. Hardware encoders need to do everything they're going to do with a frame of video in real time, where a software encoder is able to slow down as needed for complex frames.

Hardware encoders definitely had the quality edge over software in the 90's, but I'd say the best software exceeds the best hardware now. I did a bunch of SD MPEG-2 testing last fall, and this was definitely true there.

I think you are over simplifying what HW encoders can do. True, a HW encoder only gets a limited time to optimize the encode if it is for real time broadcast like a live sports event, but there are no strict constraints to maintain a stable 24 or 30 fps for offline disc encodings. And true also that SW encoding is the most flexible. But at some time or another the process becomes mature enough to cast the software into silicon and even then most times the HW part is used for the "heavy lifting" with encode optimization left to SW for the "clean up" part of the job.

b2b

I think you are over simplifying what HW encoders can do. True, a HW encoder only gets a limited time to optimize the encode if it is for real time broadcast like a live sports event, but there are no strict constraints to maintain a stable 24 or 30 fps for offline disc encodings. And true also that SW encoding is the most flexible. But at some time or another the process becomes mature enough to cast the software into silicon and even then most times the HW part is used for the "heavy lifting" with encode optimization left to SW for the "clean up" part of the job.

This hw vs sw debate is eternal. Ben was just saying currently the best sw has upperhand over best hw.

This reminds me of the unmanned robot race I saw sometime back on INHD (or was that discovery HD). To cut the story short the team that did most of the stuff in software with fewer people and lesser budget won over the other team that had a big budget and mostly hardware.

Going by these comments, PS3 should be the best player out there over the period of time since it is software based. All these SoC based players can bite the dust being hardware based. Or is it? :D

Going by these comments, PS3 should be the best player out there over the period of time since it is software based. All these SoC based players can bite the dust being hardware based. Or is it? :D
The comments were regarding encoders. Decoders are much closer to each other than encoders...

Amir

This hw vs sw debate is eternal. Ben was just saying currently the best sw has upperhand over best hw.

This reminds me of the unmanned robot race I saw sometime back on INHD (or was that discovery HD). To cut the story short the team that did most of the stuff in software with fewer people and lesser budget won over the other team that had a big budget and mostly hardware.

Don't know the HW processor setup for Sony's encoder, but Amir mentioned the VC-1 Pro encoder set-up was 8 processors and the source was split-up for parallel input workloads. So yes SW can be good, but it still depends on raw HW horsepower to give performance. I would bet someone is working on a CELL based H.264 encoder too.. ;)

b2b

So yes SW can be good, but it still depends on raw HW horsepower to give performance.

The question is not hw or sw. The question is what part be done in HW and what in SW. A judicious mix is what is really needed ... and it looks like currently it is judicious to do more in SW.

PS : In my previous lives I've been a hw engr ... so I'm not purely talking from a sw perspective.

For anyone with MPEG-2 encoder questions I would suggest that you direct your question(s) to dr1394 (Ron). He is an engineer in this area. Another very experienced poster is Glimmie. IMHO the FUD and BS posted here greatly exceeds the facts posted here!!

For anyone with MPEG-2 encoder questions I would suggest that you direct your question(s) to dr1394 (Ron). He is an engineer in this area. Another very experienced poster is Glimmie. IMHO the FUD and BS posted here greatly exceeds the facts posted here!!

It's a tough subject... They even offer it as a course of study for EE students.

http://www.ee.ryerson.ca/~gnkhan/ELE700Topics/GK03.html

b2b

For anyone with MPEG-2 encoder questions I would suggest that you direct your question(s) to dr1394 (Ron). He is an engineer in this area. Another very experienced poster is Glimmie. IMHO the FUD and BS posted here greatly exceeds the facts posted here!!
Apart from the fact that Amir works for MS and may be a little bit of PR is part of the job, I can clearly see that he loves the work they have done on VC-1. Anyone feeling that MPEG2 (or AVC for that matter) is put to disadvantage because of so many posts from him (and Ben) should jump in and clear the air instead of waiting for questions specifically directed towards them. It is an open forum.

No offense to you, Wendell. Just my few cents regarding the FUD and BS part.

Personally, I am hoping that MPEG2 VBR 18Mbps may just be fine for initial movies. But I am finding it hard to wait until movies come out. The best answer would be if someone put a hi-def clip using MPEG2 VBR. That should give us a good preview.

No offense to you, Wendell. Just my few cents regarding the FUD and BS part.None taken.
Personally, I am hoping that MPEG2 VBR 18Mbps may just be fine for initial movies. But I am finding it hard to wait until movies come out. The best answer would be if someone put a hi-def clip using MPEG2 VBR. That should give us a good preview.If you know someone with a D-VHS D-Theater tape machine then they could show you what you can expect. Blu-ray Disc can exceed D-Theater by a fair amount in the data rate department.

See the The Official AVS D-Theater Rating Topic! (http://www.avsforum.com/avs-vb/showthread.php?t=367236) for HD movie reviews. One thing should become very clear to you. HINT: It has nothing to do with encoders (or formats). You may want see this post (http://www.avsforum.com/avs-vb/showthread.php?p=7613816&&#post7613816) by me.

If you know someone with a D-VHS D-Theater tape machine then they could show you what you can expect. Blu-ray Disc can exceed D-Theater by a fair amount in the data rate department.

That is one of the reasons I am optimistic about MPEG2 VBR movies: by extrapolating. But nothing more than that. Since VC-1 is out there, and MPEG2 VBR is not, it is to the VC-1 supporter's advantage. Everytime someone tries to show by calculations and extrapolation that MPEG2 VBR 18Mbps will give close to transparancy they can be easily put down by VC-1 supporters because VC-1 is out there. I tried calculations myself here (http://www.avsforum.com/avs-vb/showthread.php?p=7592996&&#post7592996) . We need a trailer; forget about comparing with VC-1, I just want to see the quality! Where are Sony insiders? :)

If you know someone with a D-VHS D-Theater tape machine then they could show you what you can expect. Blu-ray Disc can exceed D-Theater by a fair amount in the data rate department.
Are you talking about peak rate or total file size? Because if it is the latter, your statement is not true. D-VHS has a raw bandwidth of 28.2 Mbit/sec (http://www.jvc.com/press/index.jsp?item=419&pageID=1). Take some off for data correction and such, and you wind up with 25 Mbit/sec. With audio taking 1 Mbit/sec or so, we have 24 Mbit/sec left for MPEG-2. Compared to 18 Mbit/sec that is rumored to be used by Sony on BD titles, we are talking about a file size that is 25% lower.

Yes, yes, VBR is used to encode BD as opposed to constant bit rate in D-VHS. But you don't get something for nothing. If you allocate bits to difficult segments, you have to take them out of non-difficult areas. And in those segments, you have less bits than D-VHS equiv. Far less. Imagine a difficult sequence lasting for a few seconds and BD encoder allows peaks of 24 Mbit/sec. This means that the actual data rate for adjacent frames now needs to be just 12 Mbit/sec to compensate. Meaning that these segments are running at half the data rate of D-VHS while the peaks only match it.

If you let the peaks exceed D-VHS, you make the situation much worse for the rest. Imagine peaks of 36 Mbit/sec for a few seconds now. This should outperform D-VHS. But the adjacent frames shrink down to 9 Mbit/s for double the amount of time to compensate. In such situation, it would very hard to achieve transparency. And since you have to run at the lower bit rate for so long, the artifacts will be visible.

So it is really a fallacy that VBR is better than CBR. It is not in absolute sense. It is only better if you can use VBR to exceed the average rate of the comparable file. If instead, you use it to rob the bit budget for the rest of the movie, something has to give.

I will admit of course that the skill of the operator has a lot to do with this. And movies no doubt can be made to look really good, despite the 25% reduction in file size, especially if the full rate was not needed to achieve the equiv. D-VHS quality. But one can't assume that in all situations, BD is equal or better than D-VHS. It won't be. It can't.

As an aside, note that there is no such thing as a real CBR. All CBR encodings are still variable rate within a short window. So even there, one can cover short peaks of action with higher data allocation. VBR of course lets you do this for much longer period of time. But the contrast with CBR is not as high as people think. So even D-VHS is VBR over this short window. As such, the gap is narrower than one thinks.

Let’s finish with some “math camp.” According to JVC, you can have 2.5 hours in the shortest tapes available. Using the data rate of 28.2 Mbit/sec, this translates to 31.7 Gigabytes. This is already larger than both HD DVD-30 and BD-25. And in the optical formats, we have multiple audio tracks at higher fidelity taking up a ton more space and one needs room for extras. If we don’t use advanced codecs to gain efficiency, the math simply does not work across all kinds of content, and skill/time of the operator.

Amir

D-Theater pre-recorded content is encoded without inverse telecine at 23 Mbps. If the same content is encoded with inverse telecine or at 24 fps native, then 23 * 0.8 = 18.4 Mbps will provide almost the exact same level of video quality.

Ron

D-Theater pre-recorded content is encoded without inverse telecine at 23 Mbps. If the same content is encoded with inverse telecine or at 24 fps native, then 23 * 0.8 = 18.4 Mbps will provide almost the exact same level of video quality.

Ron

That and not to mention niffy tools like this one from Sonic>Q: ExpressQC™ sounds too good to be true. How does it work?

A: Sonic CineVision automatically analyzes the PSNR (Peak Signal-to-noise Ratio) during encoding by comparing each encoded frame with the original source frame to give a value which represents how similar they are. ExpressQC rapidly searches the entire encoded stream for segments that fall below a specified threshold, or simply for the specified number of segments with the worst PSNR rating. This accelerates the QC process by instantly highlighting segments which may benefit from being re-encoded at a higher bit rate.

I have no idea if it works as advertised but clever things like this can put the bits where they are needed while maintaining better quality than the shotgun approach of CBR.

b2b

As an aside, note that there is no such thing as a real CBR. All CBR encodings are still variable rate within a short window. So even there, one can cover short peaks of action with higher data allocation. VBR of course lets you do this for much longer period of time. But the contrast with CBR is not as high as people think. So even D-VHS is VBR over this short window. As such, the gap is narrower than one thinks.
Amir

So all those years of DVD using VBR has been a waste. And all those histograms of bitrates bouncing up and down have shown us nothing. :rolleyes:

Example: http://www.dvdbeaver.com/film/DVDCompare2/crouchingtiger.htm

I see bounces from 4Mbps to 8Mbps in the standard and somewhat smaller range with the superbit.

Here's a better one T2: http://207.136.67.23/film/DVDCompare/term2.htm

And the funny thing is that the Extreme edition has a lower average bit rate, more VBR activity, but a better encode quality than the Ultimate edition as evidenced by the screen caps at the bottom of the page.

b2b

D-Theater pre-recorded content is encoded without inverse telecine at 23 Mbps. If the same content is encoded with inverse telecine or at 24 fps native, then 23 * 0.8 = 18.4 Mbps will provide almost the exact same level of video quality.

Surely they're using field_repeat tags with D-Theater with 24p source, so you'll still only be getting 48 unique fields per second, with 12 repeats.

Unless they're using BAD encoders, there shouldn't be any difference at all in data rate or image quality between 60i with 3:2 flagging and a native 24p bitstream.

So all those years of DVD using VBR has been a waste. And all those histograms of bitrates bouncing up and down have shown us nothing.
Reread his post. He's not saying that CBR > VBR. He's saying that sometimes a CBR > VBR if the CBR has a data rate advantage.

Say, BD with an average of 18 Mbps and a peak of 40 v. a D-VHS CBR 24 Mbps. It isn't clear at all which one will be worse or better. The BD would provide a more consistant quality, but it might on average be worse than the CBR, but the CBR might be worse on the hardest content (like really fast motion or a cross-dissolve).

VBR varies bitrate in order to keep quality constant
CBR varies quality in order to keep bitrate constant

From a codec perspective, the 18 Mbps VBR might have a worse average Q than the 24 Mbps CBR, even if the CBR has a worse peak Q.

For anyone with MPEG-2 encoder questions I would suggest that you direct your question(s) to dr1394 (Ron). He is an engineer in this area. Another very experienced poster is Glimmie. IMHO the FUD and BS posted here greatly exceeds the facts posted here!!

Glimmie is the man! One look at the equipment room (http://home.earthlink.net/~hdtv101/id8.html) and it's clear he knows his stuff. After looking at a few of his posts, he is definitely sharp enough to cut through the baloney post-haste.

b2b

Glimmie is the man! One look at the equipment room (http://home.earthlink.net/~hdtv101/id8.html) and it's clear he knows his stuff. After looking at a few of his posts, he is definitely sharp enough to cut through the baloney post-haste.

b2b


What a ridiculous conclusion.

Do you really think everyone that dríves e.g. a Ferrari is capable of competitive racing?

Glimmie certainly does know his stuff, but that has NOTHING to do with the "equipment room" of his home theater.

Reread his post. He's not saying that CBR > VBR. He's saying that sometimes a CBR > VBR if the CBR has a data rate advantage.

Say, BD with an average of 18 Mbps and a peak of 40 v. a D-VHS CBR 24 Mbps. It isn't clear at all which one will be worse or better. The BD would provide a more consistant quality, but it might on average be worse than the CBR, but the CBR might be worse on the hardest content (like really fast motion or a cross-dissolve).

VBR varies bitrate in order to keep quality constant
CBR varies quality in order to keep bitrate constant

From a codec perspective, the 18 Mbps VBR might have a worse average Q than the 24 Mbps CBR, even if the CBR has a worse peak Q.

I'm just going to defer to dr1394 comments about D-VHS (D-Theater) content actually being stored on the tapes in 30i format. Part of the deal with D-VHS in the first place was to store (record) OTA content direct to tape. All of the ATSC 1080 broadcasts are 30i, so doing prerecorded tapes at 24p would have just complicated to whole process just to save bits which wasn't needed given the size of the D-VHS tape bit bucket. But might be wrong on that.

Maybe that's where Sony came up with their 18Mbps for BD-25. That's a little over 16GB for a 2hr movie leaving the rest for PCM audio and what not.

They could have super clean masters that were struck with MPEG2 with a 25GB target, but I'm sure that will be fodder for other discussions at a later date. ;)

b2b

P.S. If my math is wrong, then correct me and it can be fodder too.

Surely they're using field_repeat tags with D-Theater with 24p source, so you'll still only be getting 48 unique fields per second, with 12 repeats.I thought I mentioned this when the subject came up before, but I believe "I-Robot" is encoded in a way where it makes no difference (24p source), like you describe. "U-571" (a much earlier title) looks like it might just be straight 1080i and not have that 1080p24 advantage. I could be wrong, but it also would be surprising to me if they stored "I-Robot" so inefficiently and yet intelligent devices can convert it to 1080p48 (as I have found).

--Darin

If you let the peaks exceed D-VHS, you make the situation much worse for the rest. Imagine peaks of 36 Mbit/sec for a few seconds now. This should outperform D-VHS. But the adjacent frames shrink down to 9 Mbit/s for double the amount of time to compensate. In such situation, it would very hard to achieve transparency. And since you have to run at the lower bit rate for so long, the artifacts will be visible.
Amir
Your example is just one possibility. Neither does one have to go to 9 Mbit/s to compensate for 36 Mbit/s nor does it necessarily involve adjacent frames. Where you go and when depends entirely on the characteristics of the source to be compressed. There will be sources where VBR at 18 Mbit/s will give on average worse results than 24 CBR and vice versa. The more parts of a source that can be encoded with D-Theater quality (to the human eye) at < 18 CBR the more likely VBR at 18 will outperform CBR at 24 for that source. I expect the 18 VBR to outperform 24 CBR for clean CGI titles any day. For grainy films on the other hand I don't. VBR will have to go higher on average than 18.

Your example is just one possibility.
Of course. I used that example for simplicity. The choices are infinite but the cost of making the disc is not :).

There will be sources where VBR at 18 Mbit/s will give on average worse results than 24 CBR and vice versa.
And that is the key point I was trying to make. If you agree with this, then one has to also agree that BD at 18 Mbit/sec, will not be superior to D-VHS on every movie. So we must stop using D-VHS as a reference for BD. They are different animals.

Amir

Personally, I am hoping that MPEG2 VBR 18Mbps may just be fine for initial movies. But I am finding it hard to wait until movies come out. The best answer would be if someone put a hi-def clip using MPEG2 VBR. That should give us a good preview.
I agree with this - now there's a challenge I'd like to see someone step up to.

Sadly, I suspect the response to this request is going to be deafening silence...

And that is the key point I was trying to make. If you agree with this, then one has to also agree that BD at 18 Mbit/sec, will not be superior to D-VHS on every movie. So we must stop using D-VHS as a reference for BD. They are different animals.
Amir

I hope Sony uses a better MPEG encoder than was used for D-Theater. If not and they really use 'only' 18 Mbit VBR I doubt they reach transparency compared to the D5 with material that is not squeaky clean.

I'm just going to defer to dr1394 comments about D-VHS (D-Theater) content actually being stored on the tapes in 30i format. Part of the deal with D-VHS in the first place was to store (record) OTA content direct to tape. All of the ATSC 1080 broadcasts are 30i, so doing prerecorded tapes at 24p would have just complicated to whole process just to save bits which wasn't needed given the size of the D-VHS tape bit bucket. But might be wrong on that.
You're making an innaccurate distinction between 24p encoded at 60i and native 60i. A decent MPEG-2 encoder will pick up the 24p source, and use field repeat tags as useful. Now, it might not be 100% perfect cadence, like problems seen in early "progressive" DVDs, but it'd be almost perfect as far as compression efficiency goes.

Even a bad MPEG-2 encoder will still do pretty well with 3:2 source, since 12 fields out of 60 won't have much delta, and won't require many bits even if there is a little analog noise or something that keeps a field_repeat tag from being used.

Native progressive encoding ala HD DVD or Blu-ray is very useful to enable progressive playback in devices, but is a marginal improvement at best for compression efficiency, assuming a competant MPEG-2 encoder.
Maybe that's where Sony came up with their 18Mbps for BD-25. That's a little over 16GB for a 2hr movie leaving the rest for PCM audio and what not.
I suspect came up with that figure because that's all the bits left over after PCM audio and SL discs...

One thing should become very clear to you. HINT: It has nothing to do with encoders (or formats). You may want see this post (http://www.avsforum.com/avs-vb/showthread.php?p=7613816&&#post7613816) by me.

I think we have argued this before ... with you claiming only professions should talk about it and others should shut up. So, using your logic in this case only people involved in writing encorders should talk ... :p

I have about 25 D-VHS D-Theater movies. These movies are MPEG-2 encoded with realtime encoders and CBR (Constant Bit Rate). The artifacts are few and one must look very close and carefully to find any.

We are not doing all these upgrades to hidef dvd to get artifacts - few or otherwise.

If anybody is interested this is the S&V article (http://www.soundandvisionmag.com/article.asp?section_id=2&article_id=1441&page_number=2) where Sony says they will be using 18Mbs. I assume they will be using this bit rate just for the initial releases that are on BD-25 disks?

The question really hinges on what are the "initial" releases. I think the collective understanding puts this at least through the end of 2006, and maybe longer.

What a ridiculous conclusion.

Do you really think everyone that dríves e.g. a Ferrari is capable of competitive racing?

Glimmie certainly does know his stuff, but that has NOTHING to do with the "equipment room" of his home theater.

More proper question is: does everyone that drives a Ferrari in competitive racing know how to build one?

Ridiculous snide comments welcome. I'm sure they will follow. :rolleyes:

b2b

You're making an innaccurate distinction between 24p encoded at 60i and native 60i. A decent MPEG-2 encoder will pick up the 24p source, and use field repeat tags as useful. Now, it might not be 100% perfect cadence, like problems seen in early "progressive" DVDs, but it'd be almost perfect as far as compression efficiency goes.

Even a bad MPEG-2 encoder will still do pretty well with 3:2 source, since 12 fields out of 60 won't have much delta, and won't require many bits even if there is a little analog noise or something that keeps a field_repeat tag from being used.

Native progressive encoding ala HD DVD or Blu-ray is very useful to enable progressive playback in devices, but is a marginal improvement at best for compression efficiency, assuming a competant MPEG-2 encoder.

I suspect came up with that figure because that's all the bits left over after PCM audio and SL discs...
After a long search I did find thisThe majority of movies that Hollywood is making available in the D-Theater format are mastered in 1080p then converted to 1080i before they are mastered and duplicated. The irony of this is that the majority of 1080p film transfers only have enough resolution to support the horizontal capability of 720p. Horizontal resolution of most film masters in 1080p is in the area of 800 to 1300 lines. The horizontal capability of 720p is 1280 lines, the top end of what's on the 1080p master. An argument could be made that even with the 30% vertical filtering in the 1080p to 1080i conversion the vertical resolution of 1080i is still slightly better than 720p. It's 756 lines as opposed to 720 lines. Interlaced artifacts in 1080p/24 frame material converted to 1080i/60 are not nearly as bad as video material initially created in 1080i/60.

http://shop.store.yahoo.com/dvdinternational/dve-faq.html

b2b

Video data reduction (compression) is complex. To do well requires experienced people, monitoring, and analyzing. At this point in time it appears there is ample resources for doing MPEG-2. I can find very little resources for AVC/VC-1 at the professional level.

One of the premier sources for professional test and monitoring equipment is Tektronix® corporation. Click this Tektronix® Video Test (http://www.tek.com/products/video_test/) link and have a look at some of the equipment. There many links to Tech Notes, Briefs, etc.

Video data reduction (compression) is complex. To do well requires experienced people, monitoring, and analyzing. At this point in time it appears there is ample resources for doing MPEG-2. I can find very little resources for AVC/VC-1 at the professional level.
The tools are private at this point. Fortunately the major post houses are fully trained on them and are in full production (for VC-1 at least).

One of the premier sources for professional test and monitoring equipment is Tektronix® corporation. Click this Tektronix® Video Test (http://www.tek.com/products/video_test/) link and have a look at some of the equipment. There many links to Tech Notes, Briefs, etc.
These tools are mostly for broadcast engineering. For post production, there is no replacement for a sharp eye of the compressionist.

Amir

I agree with this - now there's a challenge I'd like to see someone step up to.

Sadly, I suspect the response to this request is going to be deafening silence...
To be fair, I didn't see any VC-1 encodings before the HD DVD release either. It will be nice to have some clips, but if we don't get such clips, I can understand that too instead of spinning it other way!

I agree with this - now there's a challenge I'd like to see someone step up to.

Sadly, I suspect the response to this request is going to be deafening silence...
I have the equipment, but obtaining uncompressed source that's copyright free is the problem. It also has to be a long enough clip for the VBR algorithm to do it's thing.

In the meantime, here's a bitrate graph of a Chinese EVD movie (Happy Go Lucky) that I was using as test material while developing the EVD HD MPEG-2 encoder.

http://img143.imageshack.us/img143/4301/bits4eu.jpg (http://imageshack.us)

The vertical axis is in bits/frame, so multiply by 24 to get bits/second.

Ron

what are the red black and green lines

dr1394, thanks for stepping in! :)

Is this representative of the movie in general? I mean, are the lows and peaks and their separation/repetition that we see here represent the movie in general (roughly)? What is the red line? And does this target (22Mbps peak, 10Mbps average) provide transparency to source for this resolution? Sorry, too many questions! I trying to avoid jumping to conclusions!

And are you still using openwin? :D


EDIT: AnthonyP already asked about the lines before I posted! I was going to ask about green line later. I thought the black represented instantaneous bitrate.

As a wild guess the red and green lines are respectively short and long term moving averages of the bits/frame. The green line appears to be a simple average over about 5000 frames. The red one is too short to tell.

- Tom

Video data reduction (compression) is complex. To do well requires experienced people, monitoring, and analyzing. At this point in time it appears there is ample resources for doing MPEG-2. I can find very little resources for AVC/VC-1 at the professional level.

That is always the case with new technology. But you have to start doing stuff to get experience in it .... otherwise we would all still be using cobol on IBM mainframes.

what are the red black and green lines
The black lines are bits per frame. Since the graph contains so many frames, it's all squished together and you can't see each frame. So the top of the black area is the largest I-frames and the bottom is the smallest B-frames.

The red line is a running average over 24 or 30 frames (can't remember now), so it represents the VBR rate over 1 second (or two GOPs to get a smoother average). The peaks are just below 900000 bits/frame. 900000 * 24 = 21.6 Mbps

The green line is the running average over 5000 frames and shows the average rate (that could be used to calculate expected disk space to be used). It's a little above 400000 bits/frame. 400000 * 24 = 9.6 Mbps

dr1394, thanks for stepping in! :)

Is this representative of the movie in general? I mean, are the lows and peaks and their separation/repetition that we see here represent the movie in general (roughly)? What is the red line? And does this target (22Mbps peak, 10Mbps average) provide transparency to source for this resolution? Sorry, too many questions! I trying to avoid jumping to conclusions!

And are you still using openwin? :D

10000 frames at 24 fps is about 7 minutes. So you're seeing a trailer from frame 0 to frame 3000 or so. Then it's probably just black screen for about 300 frames. From frame 3300 to about frame 3800 is actually the same content as frames 0 to 500. Note that the bitrate is higher for the second instance. That's because this is single-pass VBR, and one of the input parameters is total frames to process. When there's still many frames to go, the bit budget is limited in order to not blow all the bits at once. The main feature starts at around frame 4100.

For 720p@24, 22 Mbps peak and 10 Mbps average worked pretty well. "Transparent" is in the eyes of the beholder. 22 Mbps was actually a little too high for EVD, which used a 2X DVD loader. I must have been trying to show that even 720p content could have some large peaks.

I'm using CDE on Solaris. The graphing tool is xvgr, which has been around for a long time. This may sound anti-MS, but I do all of my actual revenue generating work on Solaris.

Ron

If I may ask, how bit rate intensive is 1080 compared to 720? In other words, to get an equivalent picture quality at 1080P for the clip used in the example, what would the bit rate average have to be, and peak as well?

BTW, thanks for sharing your knowledge of this area with us.

Er, the license I got from Dreamworks on that does NOT allow third parties to reencode and distribute it.

Also, that clip was encoded with a MPEG-2 encoder that does adaptive softening, so it's not a particularly useful test for the D5 workflow.

ah sorry didn't knew that, here's another one

http://rapidshare.de/files/19907080/Silent_Hill-Trailer-High_Profile-5.1-LC-AAC-8000kbps.part1.rar.html
http://rapidshare.de/files/19907412/Silent_Hill-Trailer-High_Profile-5.1-LC-AAC-8000kbps.part2.rar.html

SCNR
Kolosos

More Practical Usage of Measuring Instruments for the Blu-ray Disc (http://www.blu-raydisc.com/assets/downloadablefile/2nd_bda_taiwan_usage-13130.pdf) can be found here. It is a PDF file.

For Blu-ray Disc that use MPEG-2: Video streams shall be MPEG-2 video format (ISO/IEC 13818-2) compliant. Click ISO/IEC 13818-2 (http://le-hacker.org/hacks/mpeg-drafts/is138182.pdf) for more information. This is a PDF file (255 pages). Just to be clear, they (BD) can use AVC or VC-1 as well.

Thanks, dr1394. You just made the thread more interesting.

I also want to ask what Skogan asked. Do you have any data or projection on 1080p24?

T
These tools are mostly for broadcast engineering. For post production, there is no replacement for a sharp eye of the compressionist.
Amir
Which I find somewhat strange. One would think that after all these years of codec development and mathematical modelling of human vision properties there should be some solid insights into what kind of differences between original and compressed/decompressed version of it are visible or not. Why can't there be automatic analysis of the differences and automatic fiddling with the parameters till the differences are invisible according to a working mathematical model?

Amir, can MS get permission for a section of film for this challenge?

What I was thinking is that someone get a 30 second section of film master and hand it to MS for encoding via VC1, and also to someone to do in Mpeg2.

The the compressed end products would each be released to us here on the site to compare using our own eyes.

I would suggest a target VBR of 18 mbps, since hat is what Sony has stated they chose for their releases on Mpeg 2, and is roughly what VC1 encoding has used on the first releases to date.

Hands up who would like to see a comparison, if it will help claify the debate?

I have the equipment, but obtaining uncompressed source that's copyright free is the problem. It also has to be a long enough clip for the VBR algorithm to do it's thing.

In the meantime, here's a bitrate graph of a Chinese EVD movie (Happy Go Lucky) that I was using as test material while developing the EVD HD MPEG-2 encoder.

The vertical axis is in bits/frame, so multiply by 24 to get bits/second.

Ron

Which I find somewhat strange. One would think that after all these years of codec development and mathematical modelling of human vision properties there should be some solid insights into what kind of differences between original and compressed/decompressed version of it are visible or not. Why can't there be automatic analysis of the differences and automatic fiddling with the parameters till the differences are invisible according to a working mathematical model?
Those kinds of psychovisual techniques have been used in codecs for a while now, and are a very active area of development. They help at moderate data rates, but they aren't a cure-all - no ones ever going to make visually perfect 1080p24 @ 4 Mbps, even if they hand-craft the DCTs :).

VC-1 certainly has modes where it makes decisions based on perceptual models. I can't go into details, of course.

Which I find somewhat strange. One would think that after all these years of codec development and mathematical modelling of human vision properties there should be some solid insights into what kind of differences between original and compressed/decompressed version of it are visible or not. Why can't there be automatic analysis of the differences and automatic fiddling with the parameters till the differences are invisible according to a working mathematical model?

I posted earlier (http://www.avsforum.com/avs-vb/showthread.php?p=7614284&&#post7614284) that Sonic does. I don't know if it is as you describe, but they have ExpressQC which uses PSNR to compare encoded frames against the original to flag bad ones to be re-encoded.

b2b

Amir, can MS get permission for a section of film for this challenge?

What I was thinking is that someone get a 30 second section of film master and hand it to MS for an AVC encoding via VC1, and also to someone to do in Mpeg2.

The the compressed end products would each be released to us here on the site to compare using our own eyes.

I would suggest a target VBR of 18 mbps, since hat is what Sony has stated they chose for their releases on Mpeg 2, and is roughly what VC1 encoding has used on the first releases to date.

Hands up who would like to see a comparison, if it will help claify the debate?

I am in. (You mean "...to MS for 1080p24 encoding via VC1" not AVC, right?).

16Mbps is what is used on Serenity. I think other movies have lower bit-rate. But for comparison purpose, 18Mbps is just about right.

I third that. Excellent idea if we get both encoding from the same source.
I still haven't seen HD-DVD ( I live in europe). but I compared WMV-HD clips from M$ to some Mpeg2 Recording I own and to my eyes, Mpeg2 wins in terms of clarity, lack of banding, and picture details most of the time. Thus I am curious to see if >12Mb/s VC1 encoding would change that.

I still haven't seen HD-DVD ( I live in europe). but I compared WMV-HD clips from M$ to some Mpeg2 Recording I own and to my eyes, Mpeg2 wins in terms of clarity, lack of banding, and picture details most of the time. Thus I am curious to see if >12Mb/s VC1 encoding would change that.
Well, yes. WMVHD was an old WMV9 implementation running at ~ 8 Mbps CBR. Rest assured it's a LOT better with our code enhancements and 16 Mbps VBR!

I haven't found a case yet where circa 2006 VC-1 doesn't beat MPEG-2 with the same parameters.

We'll get you guys an encoder soon (really, soon) so you can start playing with it yourself.

I am in. (You mean "...to MS for 1080p24 encoding via VC1" not AVC, right?).

16Mbps is what is used on Serenity. I think other movies have lower bit-rate. But for comparison purpose, 18Mbps is just about right.
Yes - my boo-boo - thanks for catching that! :)

Stay tune for the new VC-1 encoder this moment

but a question still comes up, will you(Microsoft codec team) improove the encoder in the future even when the final version released?

The open source AVC project got updates everyday, can you keep the VC-1 encoder neck by neck with them?

We'll get you guys an encoder soon (really, soon) so you can start playing with it yourself.

okay ;)

Ben -

You thought of putting out professionally encoded VC1 versions of the uncompressed Park Run and Calandar VQEG clips in VC1, like on DR1394's site? I think they allow distribution and comparisons.

- Tom

Stay tune for the new VC-1 encoder this moment

but a question still comes up, will you(Microsoft codec team) improove the encoder in the future even when the final version released?

The open source AVC project got updates everyday, can you keep the VC-1 encoder neck by neck with them?
Well, we won't distribute daily builds, but this is a beta, and eventually it won't be beta anymore :).

Microsoft has offered a number of backwards-compatible codec upgrades in the past. For example, the Format SDK 9.5 added improved WMV9 encoding. And we've been continually improving WMA encoding for years while maintaining decoder compatibility back to products released many, many years ago.

You thought of putting out professionally encoded VC1 versions of the uncompressed Park Run and Calandar VQEG clips in VC1, like on DR1394's site? I think they allow distribution and comparisons.

Sounds like something we could do if our legal guys like the license. Link?

Sounds like something we could do if our legal guys like the license. Link?

Ben -

You've got mail.

- Tom

Sounds like something we could do if our legal guys like the license. Link?
ftp://vqeg.its.bldrdoc.gov/HDTV/SVT_exports/

There's also some new clips at

ftp://vqeg.its.bldrdoc.gov/HDTV/SVT_MultiFormat/

but the format is 16-bit SGI.

UPDATE: The converter for sgi16 to yuv (YCbCr) is here:

http://www.ldv.ei.tum.de/media/files/homes/oelbaum/forschung/mpeg/sgi2yuv.zip

Ron

dr1394, I watched Park Run 1920x1080 clip (http://www.w6rz.net/parkrun1920_18mbps.ts) today (18Mbps).

I used the xport.exe to extract .mpv file, and watched it with Nero ShowTime. I am seeing a lot of artifacts. Since I am watching it on a laptop (Dell E1705, with 1920x1200 resolution), I paused and stepped each frame to make sure that LCD screen was not the limiting factor. The best I can describe is as part aliasing part macroblocks. I also see a bit of de-interlacing issues every now and then. Is the source interlaced?

I know, this is a demanding shot with a full horizontal pan, but I was expecting a bit more detail, and less artifacts. I am no expert, so I cannot describe this any better than this.

Is there anything else I can do? May be a different player? What is your suggestion?

PS: Thanks for the great collection of test patterns! Great work! I didn't pay attention to your video clips until trbarry mentioned it!

ftp://vqeg.its.bldrdoc.gov/HDTV/SVT_exports/

There's also some new clips at

ftp://vqeg.its.bldrdoc.gov/HDTV/SVT_MultiFormat/

but the format is 16-bit SGI.

UPDATE: The converter for sgi16 to yuv (YCbCr) is here:

http://www.ldv.ei.tum.de/media/files/homes/oelbaum/forschung/mpeg/sgi2yuv.zip


Cool. I'm downloading "SVT Exports" right now.