Bill, a word of friendly advice on this Thanksgiving day.... You answering questions on BD-J cost is a bit like answering the proverbial "have you beaten your wife lately." . Remember, no matter how you skin this cat, BD-J is going to cost substantially higher than iHD. Just the base royalties that your company charges will be 10X higher than iHD even if the rest of BD-J is free (which we know won't be the case). So no matter which way you argue this point, you lose and lose big.

I only continue this discussion because you keep bringing it up. If you stay quiet, I might let it go. Remember, it is HP and other licensees that have biggest issues with cost. Our primary beef with BD-J has always been its complexity and boatload of legacy functionality it includes which are not used or necessary for interactivity of these formats. Higher licensing cost simply adds insult to injury.

Note that in an ironic twist, the higher complexity of BD-J, and the delay in completion of its spec, may hurt you more than us! Watch as first generation BD products come to market and leave BD-J behind. Indeed, your biggest concern should not be iHD, but BD Movie mode being "good enough" and BD-J never coming to life in BD format. In HD DVD land, the “baseline” plan is the current DVD menu system which does not hold a candle to iHD. As such, studios and equipment makers consider iHD an essential component of first generation products – quite a different situation than BD.



Come on Bill. It is a holiday. Put away your anti-MS, OT animosity against us and enjoy a nice holiday turkey as I am about to do .

Happy Thanksgiving to everyone who celebrates it!

Amir

I think your primary beef with BD-J is that it comes from a big competitor of yours. Afterall, we know the past Microsoft x Java cases.

Don't worry, I don't blame you.

All this talk of IPR somehow magically working out to meet the needs of the market...perhaps we shoudl take a trip in the wayback machine to see if this holds true...

"The recent MPEG-4 licensing structure unveiled by MPEG LA is so upsetting and so unacceptable to us [that] we're committed to looking for alternatives to MPEG-4." (studio exec, March 2002)

Hmmm...how could this happen?!?

==========================

As for the whole Blu-ray is not DVD, let's not submit it to the DVD Forum tripe, it can be expressed using a simple equation: 6C > 3C. The same equation that explains +/-, DVD-A vs. SACD, etc. Thus, we saw the 1999 Sony/Philips DVR-Red project morph into the 2000 Sony/Philips/Pioneer DVR-Blue project (once Sony and Philips convinced Pioneer to dump its 0.6mm 401nm 0.6NA DL ROM project), further morphing into the 2002 BDF Blu-ray project once they had managed to roll in Hitachi's "me too" 25GB project and convinced MEI and LG to stop their projects in favor on embracing the one true 01.mm cover layer (LG abandoning its 0.3mm 400nm 0.65NA ROM project and MEI abandoning its homegrown 50GB DS/DL 410nm project developed in conjunction with NGK). The rest is history. It's all about the Benjamins.

Personally I think 3D is a long, long ways off since I have yet to sit through a 2 hour 3D movie without getting a bit of a headache. Though I could see 3D being used for a few movies simply because some directors like the idea of it I don't see 3D really catching on anytime in the next 25 years.


I too would like to see a few more tests comparing MPEG-4 AVC MP/HP and VC-1 though as noted the type of encoder used as well as the amount of energy put into the encoding does make a big difference in quality. Also something to note though is that the ATSC tests were done to determine the best advanced video codec for ATSC.


Tom, this is some interesting information since this would explain a lot in terms of why I have heard that the VC-1 SMPTE process has been controversial.


Now this is very interesting since Microsoft is quite willing to give a encoder for VC-1 to their developers, but not to SMPTE. To say the least this somewhat explains why VC-1 has low royalties since apparently you would need to be a Microsoft developer to use it fully. Tom, I can see why Amir jumped on you the second you entered this thread since I doubt that Microsoft would want this information to become common knowledge.


Well I must admit I have heard some pretty scary stories about Microsoft, but I am rather amazed they keep going after you. I mean if true this would not be borderline illegal but literally illegal.


Amir never said that the roadblocks were not based on the principles of the SMPTE organization .


Tom, just to let you know but Amir has basically been telling us over the last year that MPEG-4 AVC HP was made because of how bad MPEG-4 AVC was. To say the least I questioned him a lot of this issue and he would always go back to those DVD Forum tests to "prove" how bad MPEG-4 AVC was.


Mike, I know you tend to defend Amir but if what Tom has said is true than to be blunt VC-1 should not get SMPTE certification. The whole point of it from what Amir kept telling us was supposedly to make it an open standard, but from the sounds of it is no more of a standard than the Xbox is. If only Microsoft developers can fully use VC-1 than how the heck can Amir keep telling us that VC-1 is an open standard?


Alex, I agree with the idea that competition can help with prices but from what Tom has said VC-1 is not really an open standard in the traditional sense of the word. Also from what Tom has hinted at adding VC-1 decoding to MPEG-4 AVC chips is not as cheap as Amir has told us.


True, and in the end it is the picture quality that matters most.


Perhaps it is complicated but the idea that we will never really know the whole story is something that I would disagree with. In fact the VC-1 SMPTE certification controversy would make a rather interesting book in my opinion.


Amir, somehow I doubt that Microsoft alone caused that to happen and most likely most of the companies that made MPEG-4 AVC wanted that.


That is rather interesting but it is not unexpected that a new console will launch with a good number of bad units. That is true of almost all console launches though I am rather surprised to hear that the Xbox 360 is noisy.


True, and at the time it probably made sense to Toshiba since they had a lot invested in HD-DVD and would have gotten a good portion of the royalties if it won. Now though I think that Toshiba may have regretted that past decision not to join the BDA.


The DVD Forum was never a standards body and was mainly designed to help parcel out the royalties on the DVD format. The reason that Sony and many other companies left the DVD Forum to make the BDA is because they would have been stuck with that same royalty structure in the DVD Forum even if the new format was based mostly on their own work. The BDA is as open as the DVD Forum and as such I think they are equal in terms of being organizations.


Amir, I know you try to convince us of a lot but you must admit that is a pretty bad question. After all the DVD Forum is still working on DVD and as such no company is going to quit from it simply because they support Blu-ray.


Exactly.


Though not yet confirmed Amir says that AACS is supposedly technically done and that is why Toshiba can launch their HD-DVD player next month. Also any delay in AACS would not affect just the PS3 but also every other device requires AACS.


And yet Amir you still have no more of an idea of what it will cost than I do. Also Amir costing $2 to $5 is not exactly incredible when you take into account that DVD royalties alone add at least $15 to the price of a player.


Come now Amir your tirades against BD-J occur almost every week and was started long before Talkstr8t started posting in this thread. To say you are out to attack BD-J is almost an understatement and you have probably posted at least 100 posts attacking it.


And we are sure to see iHD being used to it's fullest when the first HD-DVD discs may be released . Amir saying that BD-J won't be used much at first is rather unsurprising when you consider the fact that few things are fully utilized when they first come out.


Well that is rather interesting and perhaps HD-DVD should have used HDMV as their baseline interactive layer ?

Richard, I must say, you wear your intentions on your sleeve. So hungry and willing to accept the word of anyone who attacks Microsoft, that you believe anything they say. Never mind that they don’t provide a fraction of the data points that I provide. What they say must be true because they back up what you want to believe, facts notwithstanding. Let me show how wrong you are (and yes, I will address Tom’s other claims soon).

Please take a look at this paper/proposal made to JVT (ISO/MPEG standardization activity for MPEG-4 AVC) by Sand Video: http://ftp3.itu.ch/av-arch/jvt-site...va/JVT-H029.doc. This is a proposal by Steve Gordon to add “ABT” to HP profile of AVC codec. What is ABT and who is Sand video?

Sand Video was a start-up making AVC encoders and decoders. They were a strong proponent of AVC as you can imagine, participating in DVD Forum codec shoot out, and doing demonstrations at shows such as IBC conference, claiming that AVC was superior to WMV-9 (VC-1). Before their products came to market, Sand Video was sold to Broadcom. Yes, Tom’s current employer! Broadcom’s AVC products come from Sand Video group.

ABT stands for Adaptive Block Transform. This is technology used in VC-1 to pick different size encoding blocks to both increase compression efficiency and preserve high-frequency/high-texture parts of the image (MPEG-2 by comparison, uses fixed 8x8 blocks). Steve Gordon of Sand Video proposed that the same technique be added to AVC, as part of HP profile to gain the same efficiencies. The layman analogy would be going from a car with one or two gears, to 5 or 6.

Now, here is the punch line. Are you sitting down Richard? If you look at the paper, you see references to a set of tests which were called Movie 1… Movie 5. Guess what? These were the DVD Forum test clips!. ("Five HD-scans of major release Hollywood movies (designated Movies 1-5)." Sand Video made this proposal shortly after catastrophic failure of all the AVC proposals, including their own encoder, in preservation of HD detail in the studio clips. They rightly deduced that by using fixed size blocks, AVC could not possibly match VC-1 in its performance and indeed would even lose to MPEG-2 (due to its strong loop filter which MPEG-2 does not even use). So they were in an extreme hurry to add the same feature to AVC.

Don’t believe me? Here is they synopsis from the paper itself



Richard, I hope this at least gets you to pause next time you so willingly accept the outlandish claims of one of the most bitter enemies of VC-1. You may not like what I am about. But I do stand behind what I say and have the facts to back them.

Amir

Been there. Done that. Our's is in October.

Although, at the risk of being a pooper, I saw one cynic refer to it as: Happy Native American Genocide Day.

Seriously, a Happy Thanksgiving to all our brothers and sisters in the US. Spend some time with your family for change, instead of us!

Gary

Since you're a proud owner of one of those hard to find babies, perhaps a trip in the wayback machine would be useful in getting some perspective why the DVD Forum was created in the first place. Was it to intended to be for all shiny disc ideas then and into the future? The name may hold a clue.



Yet, the underlying implication here is that some company or group of companies were EXPECTED to submitted whatever they came up with to a forum created to promote DVD?

My guess is the DVD Forum itself was all about the Benjamins too.

Gary

Amir, believing wrong of Microsoft is not exactly hard considering it's history both distant and recent. As for Tom and what he has said it is not hard to believe him since if he was lying Microsoft would have sued several times over by now. Also I have heard similar though less detailed information about why VC-1 has not gotten approved by SMPTE so this is not completely new info. It is only the level of detail that is surprising and fits together well with a lot of other things I have heard. Of course I am interested in what you have to say as well since I do believe in getting both sides of the story.


Amir, that is interesting though Tom has posted that his company was making products using MPEG-4 AVC.


Amir, to claim that MPEG-4 AVC can not equal MPEG-2 is pure BS and you should feel a bit bad even for claiming that. Even the Quicktime 7 videos, which Ben says use a pretty poor MPEG-4 AVC encoder, are better than MPEG-2 at an equivalent bit rate. As for MPEG-4 AVC HP I do agree completely that it was made to have enhanced quality over MPEG-4 AVC. That though is not exactly a negative from what I can see since both Blu-ray and HD-DVD have adopted MPEG-4 AVC HP. Also Amir I am not entirely sure how you can attack them for this since even VC-1 had an advanced profile made for it.


Gains in perceptual quality I can easily believe Amir but you went a good deal beyond that in what you said about MPEG-4 AVC MP.


Amir, of course I am interested to hear your side of the story.


Amir, I am not entirely sure what you mean by that since the only thing I am against is your attacks against Blu-ray, BD-J, and MPEG-4 AVC HP. I have no problem with you answering questions about HD-DVD and making a few promotional posts for VC-1/iHD. It is the attacks that I do not like Amir and many times their is little to no proof backing them up.

There are so many wars being fought in this thread it is almost impossible to keep score now. In no particular order:

Hollywood vs consumer rights

BD-J vs iHD

Microsoft vs Sony (both with their game machines)

MPEG-2 vs AVC vs VC1

Increasingly Amirm vs TomM

Computers vs dedicated CE devices

Discs vs downloads and networks

Probably patent pools vs anti-trust laws

DVD classic vs HD

BD vs HD-DVD (not necessarily & completely aligned to any of the above)

I have no idea who is winning but we have much of the tech world involved in all of this.

Happy Thanksgiving all. Don't care if you celebrate it, be happy anyway.

- Tom

There seem to be a bit of confusion (if not Ad Hominem) here. There was no 8 bit 4:2:0 HP at the time, that came much, much later, and it almost didn't happen at all.

ABT was part of the H.264/AVC draft standard until October of 2002, when it got thrown out during the Geneva meeting because it wasn't working right and we needed to move forward. ABT was strongly supported by HHI and others at the time. Long, long before Sand or Broadcom got involved.

We (JVT & MPEG & VCEG) issued a Call for Proposals ("CfP") to get Professional Profile (later FRExt, much later High Profile(s)) in March of 2003. As mentioned in prior Email, the justification for this CfP was to add 10 & 12 bit sample depths and 4:2:2 & 4:4:4 chroma sampling (and also to study support for RGB etc instead of just YCbCr). There was no ABT or 8x8 in the plan, and there was certainly no new 4:2:0 8 bit Profile in that plan.

Take a look at Steve's contribution. That wasn't until a couple of months after Professional Profile work was approved by our parent parties and we actually embarked on the work. If you really want to do your forensic history analysis, you'll discover that Steve's proposal was very poorly supported because we'd just thrown out ABT 6 months earlier. It wasn't until September of 2003 at the San Diego meeting that Steve was able to garner support from enough other companies for this to get serious consideration as an addition to FRExt (which I think became the new name at that same meeting).

Finally, we didn't even make up an 8 bit 4:2:0 High Profile (which is the High Profile under discussion here) using the 8x8 transform until July of 2004. What we ended up over a year later was ***NOT*** ABT. It was only the addition of an 8x8 transform to the Standard (incidentally, variations on the 8x8 transform had been proposed by FastVDO and others well before ABT itself, in January of 2002 I believe).

Not sure why I am wasting my time here trying to make sure that history is not re-written.... This will probably be the last time I respond to any of Amir's EMail. Life's too short.

As probably shown by some of my other posts in this huge thread I get confused by which additions for HD are in High Profile only . Is the 8x8 transform for 8 bit 4:2:0 intended to be in consumer AVC HD discs or is that only a FrExt studio thing?

And sorry about ABT not getting there.

- Tom

I hate to break it to you but the cost of UMD right now is significantly higher than DVD. This is especially true for lower quantity runs.

For a 5000 unit run on DVD with packaging the cost is less then 50 cents per unit. For a 1000 unit run of UMD it is more than $2. People generally do not run 1000 units of DVD so I really can't give you a quote on that. Also except for large studios they don't do 5000 unit runs of UMD.

Doug

If ABT is used in VC-1 and it was Steve from Sand Video making the proposal, then who owns the IP rights to ABT? Sand (now Broadcom) or Microsoft or some other company? That would be the more interesting question, who developed it.

b2b

Maybe this one?? Eastman Kodak owns it..


http://freepatentsonline.com/4774574.html

The IP in Java is well-known (the way the Java Community Process works requires companies contributing to the spec to make their IP known and available as well), but there could be new IP in BD-J which has not yet been declared. At the same time, it's absolutely possible that iHD is infringing someone else's IP, just as Microsoft has been judged to have infringed Eolas Technologies' IP with Internet Explorer and has been ordered to pay Eolas over $500M+ in damages.
Thank you for making my point! The initial terms were poorly received by industry and it was clear that MPEG-4 would see limited adoption based on those terms. The IP holders realized they needed to reduce costs, renegotiated, and later in 2002 announced new terms which have been well-received. This exactly demonstrates that unreasonable patent terms don't stick, they get revised until they are reasonable.

It's not OT or anti-MS. MS makes it's cash from selling OS's and Office. Almost everything else is done for free or nearly free to support that business. AND THIS IS FINE.

The ony reason you are involved in the HD disc debate at all is because MMC will allow you to sell a new generation of operating sytstems - make money.


If in the process, I as a consumer get a kick-ass HD format that is well supported, then great!

If instead, I as a consumer get some crippled, poorly built format or one that has no movies on it, then boo!

Which is the only reason why people question your motives for being here and the statements you make. Your company has a vested interest and to belive it is also, just coincidentally, the rosiest in EVERY respect for the conusmer is very hard to swallow.


As long as you post here representing MS, then nothing about MS is going to be off topic.



And happy Turkey Day!

Do you really want to celebrate a patent pool coming out with fees so high as to upset some of its top customers (studios)? Should we expect the same with BD-J?

Can you provide any assurance that unreasonable fees ALWAYS come down? If so, please tell me why OMA DRM fees have stayed at $1/device despite the huge uprising against it.


Well received where? How many MPEG-4 part-2 design wins do you see? It is not part of these next gen formats now, is it? MPEG-4 part 2 came and went.

Besides, what you are missing is that BD-J is currently mandatory in BD. What is Sony to do if BD-J costs too much? Stop shipment of PS3 while the price gets re-negotiated? How about Dell?

With MPEG-4 part 2, there were other choices in the market such as WMV-9, Real's, etc. Customers could simply use these other technologies and not take an MPEG-4 license. The same freedom does not exist here. Of course, if you are saying that BD-J will be changed to optional in BD and people can leave it behind if it costs too much, and use something else like iHD, then we are talking! .


Again, you seem to be a fan of "unreasonable patent terms" for the first round. This is easy for you to say since your company isn't going to ship anything. But don’t you have any sympathy for other companies who have to ship BD products while the "unreasonableness" gets dialed out of BD-J? Seems not. Thank heavens we are in HD DVD camp .

Amir

Sigh... Was there any celebration there? Did I actually say there was anything good about a patent pool starting too high? Of course not. The point, which seems to be utterly lost on you, is that for all its flaws, the patent process generally results in royalties in-line with value.
Since you're the one claiming BD-J patent fees will be so onerous, the burden is on you to point out a technology which has failed due to unreasonable patent fees, as you're claiming BD will due to BD-J.
I have no familiarity with OMA, but a quick search turns up some interesting points:

1. Terms are still being negotiated, given a recent proposal to reduce the royalty to $0.65/device and a per-subscriber charge, adding still further credibility to my argument that patent pool fees work themselves out until they are appropriate; and
2. Microsoft is one of the patent pool members (via ContentGuard), so it could in fact be Microsoft advocating for what you call an unreasonable fee.

Work within the patent pool to lower the fees to a reasonable amount, just as happens in dozens of other instances. There are probably thousands of patents which need to be licensed in order to build Blu-ray, HD-DVD, or any other comparable device. It is simply absurd to suggest that BD-J will bring things to a crashing halt while all the other patent pool technologies will be no problem. But just keep on painting BD-J patents as a real threat to Blu-ray, Amir, maybe no one will notice all the much larger, far more real threats which iHD ("brought to you by Microsoft's good will!!!") presents.

-Talk

Hi Tom. Don't be sorry about ABT. That's the way it goes in most standards development processes (in ISO/ITU, but not in SMPTE). People make proposals. We test them. Everyone tweaks and tunes and measures (and verifies each others' test results). Things that work well and have support from participants stay in the mix. Things that aren't working so well or don't have much support get tossed. Same with things that are overly complex or impractical to implement in terms of processing cost or memory - they get saved for future standards when Moore's Law might be able to make them real.

During the Professional Profile development process (later FRExt, later the High Profiles(s)), a number of things were proposed and many showed promise. But first and foremost, extending the underlying machinery to support 4:4:4 and 4:2:2 was an essential part of that program. Ditto extending the H.264/AVC codec design to support 10 bits and 12 bits of sample depth. These, in turn, required a significant re-write of the standard text (a new draft was forked off), changes in such things as CAVLC, and a fair amount of testing to make sure we got it right (and of course, changes to the reference software too so people could actually openly test these things and compare results). The natural thing to test was high definition video content and film content. (Where you do need 4:2:2 and 4:4:4 and you run at 10 bits or more, and sometimes in RGB mode (or, it turned out in the and, XYZ mode)). Note: the thinking here at the time was very much studio type applications, broadcast backhaul and contribution, D-Cinema, professional VTR replacements, etc. (That's why we tagged this amendment Professional Profile as the working name at the start, but it later became apparent that we were working on several different facets of extension, hence "FRExt" (thanks to Microsoft's Gary Sullivan who coined that term).)

It was only much, much later in the process that we start to group these things into real, documented "Profiles" for the final version of the Amendment. I guess one way to think about this would be like the way an oil refinery works - you put in the raw stock under high heat and (in this case, political) pressure and then you pick it off the cracking tower depending on your market needs. As I mentioned earlier, it wasn't until much later - July 2004 - that we had the Profile showdown in Redmond. That's where the High 4:4:4, High 4:2:2, High 4:2:0 10 bit and High 4:2:0 8 bit Profiles actually came into being. See my earlier Email on that profile discussion.

But to answer your specific question, the new tools that had been added during the Professional/FRExt, High H.264/AVC Amendment process showed enough promise through all of our content bashing that a 4:2:0 8 Bit Profile *was* picked off and set aside as a distinct profile. It is that profile which is now referenced by DVB, HD-DVD, BD-ROM and perhaps other standards bodies (ATSC? Australia's new HD system?) and network operators (DIRECTV, DISH, Sky) going forward. I don't expect Main profile to have any more design wins. (Personally, my main interest was 4:2:0 10 Bit Profile for these applications, but we apparently aren't quite there yet...)

So why don't you talk up your camp? Instead we get this incessant campaign of FUD. Give it a rest... You're starting to sound like a used car salesman trying to close a deal.

b2b

In October of 2002 the following companies had JVT contributions on ABT:

Aachen University (Germany), Sony Corp., RealNetworks, Nokia, DoCoMo Communications Laboratories USA, Inc., Videolocus (later to become LSI Logic company), Panasonic Singapore Laboratories Pte. Ltd., Sung Kyun Kwan University, Samsung Electronics., Motorola, Inc., FastVDO LLC, Sony Corp., NTT DoCoMo, Inc. (Japan)

In the July 2002 meeting @ Klagenfurt the ABT contributors were:

Heinrich-Hertz-Institute (HHI), Motorola, Inc., BCS, Aachen University.

I could go back down the history trail but this is a pretty good cross section for the purposes of this discussion. So, as I said earlier, Steve/Sand did not originate the ABT tool usage H.264/AVC. There were a number of other companies working on ABT much earlier. What happened for High Profiles was that Steve's contribution/proposal in May of 2003 to take another look at ABT was eventually transmogrified into the adoption of an 8x8 transform (not ABT) - for use in *ALL* High Profiles, not just High 4:2:0 8 Bit.

Them's the facts.

Better watch what you wish for. By reading your comment then any company with a vested interest in the next gen format is also open for those type posts.

Tom, are you a member of ISO (International Organisation for Standardisation), SMPTE (Society of Motion Picture and Television Engineers) and ITU (International Triathalon Union ? If so, who do you designate to do your sleeping for you? And which do you prefer, swimming, cycling, or running?

As I understand it both HD-DVD and Blu Ray have added MPEG4 to their specs, but just what flavour are they bound and gagged to? Main Profile, or the appropriate version of High Profile, or maybe both? Since there have been so many companies involved in it's creation is there anyone designated to try and sell this codec's use to the studios the way Microsoft seems to be biggin' up VC-1?

Also, who are the companies working on building professional encoders for MPEG4? I seem to recall hearing that Nero and a few other consumer and prosumer oriented companies were dipping their toes in the water, but what about encoders of the level studios might feel comfortable using? Is there anything up to that standard available or nearly available?

Perhaps a return to reality would serve the thread well...

1. Pioneer cited "a delay in licensing" AACS as the reason the units they are demoing this month in Taiwan have no copy protection onboard...they further noted that it will be Q1 '06 before they are cleared to ship units with AACS onboard. Hmmm...think MEI will say the same thing, or just ship their stuff (they are moving into production as we speak!) and hope no one notices?

2. Sigma Design's CC had some salient points to ponder: the SMP8634 (AVC HP/VC-1/MPEG-2 SoC) family and related ICs will be 40% of rev H2 '06, but they don't expect volume sales before 12/06...next gen DVD systems won't appear until mid '06, SMP8634 sampling got pushed back a month, but they expect to produce 10,000+ by January.

Yes (specifically MPEG), Yes and Yes (specifically ITU-T's VCEG). And BDA's TEG2 and DVD Forum's WG01. Among others.

I prefer United Airlines.



I am very uncomfortable talking about the inner workings of the DVD Forum and BDA (in fact I simply won't talk about any of that). But it is public information that they both reference H.264/AVC 4:2:0 8 Bit High Profile in their respective HD-DVD and BD-ROM specs.

One of the key marching orders when we embarked on "Professional Profile" was that it be downward compatible with other existing Profiles (such as Main) of H.264/AVC. So H.264/AVC Main Profile streams will work with a High Profile decoder. In other words, the answer to your questions above is "both".



No. Nobody is in charge of marketing H.264/AVC. But as you point out there are many, many stakeholders both technical and business-wise. Also, the Studios are smart, and they're big kids. Many of the major studios have broadcast, cable or satellite arms and responsibilities inside of these companies overlap across broadcast, motion pictures, DVDs etc. So they are all aware of H.264/AVC for a variety of reasons. In fact, for the development of Professional Profile we had input from two major Hollywood studios on the 4:2:2 tools (they wanted to make sure what we were doing would give them exactly what they needed for future high quality and more efficient broadcast applications). We had input from a number of them on the 4:4:4 RGB and XYZ tools.

Going to a tradeshow like IBC it is basically impossible to miss H.264/AVC on the floor, and all of the Studios have delegations to the major tradeshows (and venues like the HPA Tech Retreat).

Also, one of the major Hollywood authoring companies (a name you know very, very well), was a major participant/contributor in the development of H.264/AVC, and especially the High Profiles. (And Yes, they have IP in this.)



You ask three questions. Best-effort answers (without violating NDAs):

1) Yes, in fact some are already deployed at both Studio and authoring facility sites in and around Hollywood and Japan.

2) I can't talk about other people's unannounced product plans.

3) Yes - and already deployed at both Studio and authoring facility sites in and around Hollywood and Japan.

they both use the same one plainest HP, 4:2:0/8

I don't believe I ever said that Apple's is pretty poor (and I don't believe that's true). I may have said it was pretty limited (Baseline and Main profiles only, progressive only). A lot of folks talk about Apple's codec like it's crap, but I think it's pretty good for HD as of 7.0.3.

I don't believe Amir ever said that MPEG-2 is intrinsically better than AVC, either. He did point out that MPEG-2 beat AVC in some tests, which it did, but that was before HP.



Oh, I really, really hope that everyone can disagree with that! I'd personally like to see this thread get back to some details about things that'll actually matter to consumers, content producers, or CE companies. There's been way more heat than light here lately, and broad, vague speculation about stuff that posters really don't know anything about other than their own misinterpretation of carefully worded comments here is getting irritating and dull.

I've been trying hard to hold my virtual tongue on these political issues. As a SMPTE member auditing the c24 group, I know exactly what Tom and Amir are talking about, but despair about the possibility that actually sharing my thoughts on it would lead to anything else other than more senseless arguing about misunderstood details.

I'd much rather talk about the sausages, or at least the sausage factory, rather than the personalities of the sausage factory's bank's board of directors.



Tom, I'd love to hear your take on why you think we aren't quite there yet. Increased expense in the STB?

(Context here is 10 bit distribution to the consumer.)

The main instant come-back claims and opinions (which I do not share) seem to be:

1) You can't see any more than 8 bits on consumer displays.

2) Under normal consumer (living room) lighting conditions it would be pointless.

3) It would raise the bitrate in the distribution channel (this has been completely debunked - this was just folklore).

I've done enough compression, HD and D-Cinema testing (much of which has happened on consumer DLPs and Plasmas and CRTs) to know that I can easily see the difference between 8 and 10 bits if you pay attention to detail in the authoring and compression process. I know others can too. (I can *certainly* see the difference between 8 bits and 10 bits on professional D-Cinema projectors - almost everyone can.)

Next, if you look at what goes in on mastering high value content for consumer distribution:

1) It all starts as 10 or 12 bit (or even 14 bit, if linear) RGB in the telecine or HD camera, then often gets converted to 10 bit YCbCr (a round then 3x3 matrix operation), then chroma gets subsampled to 10 bit YCbCr 4:2:2 (FIR filters), then it gets converted to 8 bit 4:2:0 through additional filtering and rounding processes (more rounding and filters both vertical and horizontal), and then it gets compressed and sent down the channel (or printed on the disk in this case).

2) The reverse steps happens at the consumer end (upsample from 4:2:0, 3x3 matrices from YCbCr to RGB, etc), eventually resulting in modulation of RGB light to your eye.

Each and every one of those mathematical processes results in rounding (or worse - truncation - in cheap/dumb implementations). In a signal processing sense, that's noise and decreased SNR. Once you're into 8 bit land, and in a space where people sometimes don't pay proper attention to their gamma math either, each of those rounding/truncation operations really starts to hurt.

Keeping things 10 bits as long as possible into the first stages of the pipeline can benefit image quality - keeping 10 bits right to the display modulator is even better. (Even keeping things 10 bits deep and as long as possible into an encoder thats only outputting 8 bit streams can help bitrate and quality!).

I am not sure I have any hope for plasmas in the 10 bit sense, but I have talked with enough LCD and DLP display designers and manufacturers to know that they could definitely use more than 8 bits of pixel depth. I believe FED displays will be in the same category - 10 bits is better.

But that would require the addition of 25% more RAM to the video decoder at the CE end (and a corresponding hit in memory bandwidth, which is already high for HD). I can't deny those two points, but I do think the cost will be worth it at some point. I only wish that that point had been this year or next, not a few years out.

In any case, this issue is now being discussed in one major broadcast-related standards body. The theory is that many countries haven't deployed HDTV yet, and if they're going to do so, why not take a bigger step than what's been done here in the US (and of course have bragging rights (if not demonstrable quality)over other content distribution channels).

I don't think we've seen the end of this yet, which is why I am glad we kept High 4:2:0 10 Bits in the H.264/AVC Amendment.

(BTW - For progressive (movie) content, 4:2:0 10 Bits works just as well as 4:2:2 10 Bits, but 4:2:2 would be OK as well (it actually simplifies some of the filtering processes.)

Memory bandwidth hit takes a little bit more than anticipated since if things don't fall on nice 8-bit boundaries (assuming you use PC DRAM for lowest cost), you end up fetching stuff you don't need.

The post processing circuitry would also want to be increased to 10-12 bits (depending on where you are in the processing pipeline), increasing chip cost.

Agree.