quote:

---

Is IMAX the next New Coke?
James Hyder, LF Examiner

... In September 2008, Richard Gelfond, co-CEO of Imax Corporation, told members of the Giant Screen Cinema Association that “we don’t think of [IMAX] as the giant screen.” Rather, he said, “it is the best immersive experience on the planet.”

The company takes this position because it has chosen not to differentiate its new digital projection system in any way from the 15/70 film systems it has been installing in giant-screen theaters since 1970. This despite the fact that, according to Imax VP Larry O’Reilly, its two major digital partners, AMC Entertainment and Regal Entertainment Group, both originally wanted to brand the new screens as “IMAX Digital.” And based on the reaction Gelfond’s announcement received in New York (and on many conversations I’ve had since) many, if not most, institutional IMAX operators would prefer this as well. In short, virtually all of Imax’s customers and partners would like to see a distinct new identity for the digital system.

But Gelfond flatly rejected this possibility, offering an absurdly flawed analogy with BMW automobiles. He said that the German carmaker offers the 7-series line of larger, more powerful, luxury models as well as the smaller, entry-level 3-series cars. “People don’t say ‘The 3 isn’t a real BMW because it’s smaller.’”

Of course, this ignores the fact that the model numbers, to say nothing of the prices, clearly distinguish BMW’s different product lines in consumers’ minds, while maintaining the unity of the brand. No car buyer believes he has bought a $125,000 760Li only to receive a $30,000 328i. ...

---

quote:

---

Under the terms of the joint venture agreement, IMAX and Regal Cinemas will share the cost and profitability of the new theatres.

---

quote: James B Gardiner

---

Sony have some technical issues (4k made from 4x2k DMDs that do not seem to be able to archive the same colour)

---

I did not follow you there. 4x2k DMD's? Who is doing that?

All DCI "certified" projectors I'm aware of either use 3x2K DMD's to do 2K (NEC, Christie, Barco) or 3x4K LCos to do 4K (Sony).


 

---

Posted by James B Gardiner (Member # 5151) on 03-15-2009, 07:23 AM:
 
Julio,
from what I have been told, and it makes sense, the sony DMD, to archive 4k, use 4x2k wafers butted up together. This is what I have heard, and it would also back up the other info I heard about Sony projectors. They cannot archive consistent (To DCI spec) colour across the image. Which the above implementation would be the result of.
Ie, Can you get two TV sets, exactly the same. (No)
Try to get them to archive the same colour and gamma curve.
Ie same issue with 4x2k DMD butted up against each other. You archive the 4k Image, but also have colour consistency problems.
 

---

Posted by Julio Roberto (Member # 4976) on 03-15-2009, 08:09 AM:
 
The Sony DCI projectors do not use DMD's (Digital Micromirror Devices) at all.

They are a type of 3xLCos (Liquid cristal on silicon) panel design, which Sony calls SRXD (Silicon Reflective X-tal Display). Instead of chips made to behave like millions of tiny mirrors that can move really fast, they are made out of chips in which tiny cells of a liquid crystal material changes the amount of light reflected from them by changing the rotation of the polarization of the light coming out, which is them filtered based on this polarization variation pre-and-post reflection.

http://pro.sony.com/bbsccms/ext/SXRD/cinema.shtml
[Link is to sony's page on their cinema 4K projectors]

http://pro.sony.com/bbsccms/ext/SXRD/pdfs/SXRD_VSAP_WP_v3.1.pdf
[Link is to Sony's white paper on SXRD technology]

Therefore, all the rest of the talk about color, etc, related to some theoretical 4xDMD panel projection design doesn't apply , not to mention that I don't think it's correct to begin with.

Sony's projectors or technology doesn't have an inherent problem with color fidelity or gamma uniformity in production. They do have manufacturing difficulties and other issues which Sony has not, in my opinion, "mastered" yet, but they are not "inherent" in the technology, except for your usual case of engineering and manufacturing capabilities limits on such "delicate" technology.

From my point of view, TI's micromirrors are a more "solid" and "mature" technology (i.e. all engineering quirks have been ironed out). But it's also a less "perfect" technology.

THEORETICALLY, Lcos can surpass DLP's in just about any quality measure (i.e. contrast, color gamuth, resolution, fill factor, etc). Only in raw light level, and possibly in absolute black performance, Lcos present a THEORETICAL dissadvantage over DMD.

In practice, most Lcos technologies are hard to make right and are not as polished an straight forward as DMD's, not to mention more expensive, and as a result DLP's fear quite well against them in PRACTICAL terms, even surpassing some real-life practical implementations.

In DCinema applications even more so, where the huge light levels that are sometimes required, play against one of the weaknesses of Lcos (more fragile to high temperatures) and in favor of the main advantage of DMD's (high reflectivity, virtually no absortion of light's energy).
 

---

Posted by James B Gardiner (Member # 5151) on 03-15-2009, 08:40 AM:
 
Julio,
Sorry I used the DMD acronym to represent the CHIP, be it Mirrors or a type of Sutter over a mirror.

Still looking at your documents, the PDF for example. The image on page 13, for example shows 4 separate data paths going onto the chip. This would indicate that the CHIP may be one big wafer, but that it may also be 4 x 2K engines butted against each other on the chip. Ie having, on chip, different voltage regulators for each 1/4. And as such, resulting in the problems as described above.
The document does also mentioned how you can drive a image to the 4 sections of the chip etc.

Maybe the next revision of this design from Sony, is a revamp of the Chip resulting in overcoming the DCI issues the TI people like to point out.
I am just speculating, but I know Sony does have some technical challenges with archiving DCI specs. But then again, it does not mean it does not looks amazing on screen. The issues may not be perceivable to most people if at all.

I have seen a Sony on screen, and it looked great, but have not had test images and analyzer etc.

I would also like to point out that LCos as a domestic product has completely failed, from my understanding, with all product being pulled from market. (They had a MASSIVE fail rate in field and cost Sony a packet on warrently return from what I heard) Can anyone back this up? I must admit, you don;t hear about Lcos projectors any more.

James
 

---

Posted by Julio Roberto (Member # 4976) on 03-15-2009, 09:47 AM:
 
There are plenty of Lcos projectors on the market, with new ones recently introduced, specially from JVC.

Also, a new big player, Canon, has entered the Lcos projection market, with several models in the stores.

Lcos is not a "domestic technology" anymore than DLP is, as it's used for the highest definition projectors ever, including the sony DCinema's or the previous generation Dcinema by Huges-JVC, or the current 8K Super Hivision projectors.

Lcos is but a generic term used for all reflective liquid crystal technologies. Different manufacturer's call them different things, such as D-ILA JVC or Sony's SRXD. All just variations of the same thing and a way to differenciate between them and other image projection technologies such as transmissive LCD, GLV or DLP.

I don't know where you are getting the "4 chips". The need for four beam splitters is only due to the light having to be previously polarized. Look at page 25 from Sony's white paper and you'll see the usual simplified light path. White light is divided in three primary colors, reflected off the chips and recombined for the output. Same as DMD in essence, although in this case polarization is used as a necessity.

http://en.wikipedia.org/wiki/LCOS
[Link to wikipedia on Lcos showing a simplified light path]

quote:

---

LCoS technology has the potential to enable the manufacture of big-screen high-definition televisions with very high picture quality at relatively low cost. However, LCoS, while conceptually straightforward, can be a difficult technology to master; a number of companies have dropped out of the LCoS business in recent years. Nonetheless, as of June 2006, proprietary methods for mass-producing LCoS developed, and at least four manufacturers now produce LCoS-based rear-projection televisions for the consumer market.

Commercial implementations of LCoS technology include: Sony's SXRD (Silicon X-tal Reflective Display) and JVC's D-ILA (Digital Direct Drive Image Light Amplifier). Every company which produces and markets LCoS rear-projection televisions uses three-panel LCoS technology,[citation needed]. Sony and JVC both produce and market front-projection displays that use three LCoS panels.

Developers and manufacturers who have left the LCoS microimaging market include: Philips, Microdisplay Corporation, Spatialight, Syntax-Brillian.

Display system architectures
There are two broad categories of LCoS displays: three-panel and single-panel. In three-panel designs, there is one display chip per color, and the images are combined optically.

Three-panel designs
In a DLP device the light is separated into three components and then combined back: Two beam splitters are needed. In LCoS devices the light is additionally polarized and then analyzed; four beam splitters are needed.

---

There is no "4 chips" or anything like that going on Sony's projector at all, be DMD or any other type of device. 3 imaging chips, each receiving one color of the light splitted in 3 main paths. The additional path opened in some designs is to allow for the pre-polarization and post-absortion.

There is no "four sections" or "four imagers" or "four-anything" in this technologies that would yield any such potential problems you are describing. Even if there were, the issues of non-uniformity etc you are referring to would not be a real life issue at all.

You can calibrate two entirely separate (digital) projectors to match almost perfectly, so much more you could match "two projectors heads combined into the same machine" if such thing was necessary for some reason.

I have the feeling whoever told you about the "matching 4 imagers" didn't have a clue as to what he was saying or somehow something got lost in the translation. The four drivers paths to the chip are there because it's not practical to create such a large flat cable from such a small device. But claiming this would produce color uniformity issues would be like believing that two RAM memory modules split into two slots would yield a different result than 1 single module of twice the capacity. If you want to use all addressing lines at the same time, you would need twice the pins on the single module, which would not be practical. And regarless, the RAM would always produce the same (practical) result.

Even if internally the imager was made of 4 chips, which I don't believe they are, they could be from the same wafer even. Each wafer produces over 200 working chips, so no problem on picking 4. Even if they were from different wafers, semiconducting manufacturing capabilities have such small tolerances that they all would probably be a perfect match to begin with and not even any adjusting would be required.

Thinking that tolerances are so wide between chips in the same wafer to produce performance differences, and worse yet, significant and uncorrectable performance differences is the same as thinking that parts of one single imager chip perform differently than other parts of it.

Search on google images for D-ILA and you'll see JVC 8Kx4K imaging LCos chip. It only uses two flat ribbons cables, if that makes you feel better

Sony's problems with their SRXD stuff comes from the material used in one of his panels on their optical head not being able to stand the intended use, and degrading over time, leaving the colors from sections of that panel screwed.

Sony thought they could outdo other Lcos designs with their SXRD, but as usual over-estimated themselves and now are paying the price for putting out a technology that, "awesome" as it is, wasn't fully tested. Others have been developing for years and years, much longer than Sony, only to give up as it was too expensive to get it right. Sony came and thought it could outdo them all in no time, but was wrong.
 

---

Posted by Mark Gulbrandsen (Member # 72) on 03-15-2009, 10:41 AM:
 
Yes Julio,

The Sony LCOS chips are basically 4-2K chips parked next to each other on one common substrate!! A son'y tech person at Showeast confirmed this for me severl years ago. The assembly is fed from 4 data sources. Just because T.I. hasn't been able to reliable make chips over 2K do you think Sony can? All Son'y's doemn to get around it is park 4 of them in one place!

quote: Julio Roberto

---

Each wafer produces over 200 working chips

---

quote: Julio Roberto

---

you'll see JVC 8Kx4K imaging LCos chip. It only uses two flat ribbons cables, if that makes you feel better

---

quote:

---

Sony assembles the IMITO glass and silicon backplane prior to cutting the wafer into individual panels (singlation). In this way, Sony achieves three key features: a narrow cell gap, dust-free process and no LC gap spacers in the image area.

Sony produces the SXRD panel's silicon backplane at our Kokubu
Technology Center. The backplane uses a 0.35 micron process and an innovative drive circuit. A special planarization process yields an extremely flat silicon backplane—important for cell gap uniformity. The panels themselves are assembled at Sony's Kumamoto Technology Center. There, Sony affixes the
Index Matching Indium-Tin-Oxide (IMITO) glass to the backplane before we cut the wafer into individual panels. This process minimizes the Liquid Crystal cell gap, minimizes dust and keeps LC gap spacers clear of the image area—all important considerations in picture quality.

---

...as the do the singlation on a per panel basis, they say.

Well, even if Sony 4K SRXD panels are made of 4 independent chips (I know JVC's 4K are a single one, although perhaps their 8K x 4K is a stacked design), it still holds. That should be a non-issue whatsoever unless you think that variations on different parts of the same chip are also a problem.

For pits sake. We are even advocating using entirely two separate projectors for 3D and match their images and here we are complaining about potential differences between 4 quadrant sections of the SAME IMAGING panel of 3 panel projectors!!!

On a REALLY good day they could get over 250 working chips from the wafer Who knows what their yields are nowadays, but it's reasonable to think they could get at least 20 working chips from the same wafer in a really-really bad day. Putting 4 of those together sounds to me like a reasonable way to reduce manufacturing costs, so I'm all for it.

TI's problem is not making 4K chips, which I'm sure Sony can (and JVC can for sure). Sony's chips would be about 1/4 the size of TI's at the same resolution, so it takes Sony's 4 times less "problem" to make it. As you know, manufacturing yield goes down with the square size of the chip. Lcos is FAR smaller than DMD, by about 4 times.

If you think that manufacturing variations along say 1.2" area are "significant" ...then LCos is far superior to DLP as they can do 2K in about 0.6" in a single chip.

If TI stacked 4 DMD chips to make a 4K chip it would result on an almost 2" DLP design. Go find a rod integrator, prism light engine and lens that can do that for cheap. Sony does it in 1.55". JVC in 1.27". There are other design issue that complicate DLP scalation with size (illumination angles).

About ribbon cables, it's actually "the other way around" from what some of you guys may be thinking.

2K SRXD or DLP chips abound. Say tomorrow they wanted to introduce a new 4K product (even say a true single chip, no stacked design). The best way to talk to so many pixels is divide the electronic driver into 4 2K sections and drive them at the same time. That way you could re-use the same standard cabling and some of the electronics etc as your established 2K design.

Same thing with RAM memory etc. If you wanted to double the bus size on some design for cheap, it's better just to use the current established standard conectors, number of pins and sockets and just use two of them at the same time rather than trying to design a whole new standard with tons of pins and new "circuits", cables and interconnects unless it was necessary.

[ 03-15-2009, 02:15 PM: Message edited by: Julio Roberto ]
 

---

Posted by James B Gardiner (Member # 5151) on 03-15-2009, 08:14 PM:
 
Mark and Julio,
Thanks for the input (And Julio for the LONG osts), it is all very interesting.

Cannot wait to see Sony's new version of if projector tech.
Will probably meat DCI requirements in many ways. Real 4K path for cinema. (Still, the Lenses on those monsters... Ouch)

And the fact that the DCI players would all have to be replaced, and 4K units are only betas at best right now.
Plus no one really makes 4K in Post.. um.. we have a long way to go across the board.

James
 

---

Posted by Julio Roberto (Member # 4976) on 03-15-2009, 10:03 PM:
 
DCI *mandates* all certified servers to be able to decode 2K-24fps, 2K-48fps and 4K-24. It's not optional. All servers MUST support it.

The projectors (or perhaps servers) are allowed to scale it down to 2K, though, for putting it on the screen.

4K is not optional in (true) DCI equipment other than projectors. It must be supported or it's not DCI approved. At this moment, studios are allowing transitional non-DCI approved equipment as they know that 4K projection is not widespread yet (and probably won't for a long while), so no need to rush manufacturers with 4K support temporarily.

quote: James B Gardiner

---

Plus no one really makes 4K in Post.. um.. we have a long way to go across the board.

---

quote: Mark Gulbrandsen

---

Still in my experience 4K>2K looks better then a direct 2K. The image has a much more velvety look to it.

---

Probably what's happening is that 2K is hardly ever "true 2K" and 4K is hardly ever "true 4K".

Obviously if a film was "true 2K" (i.e. absolutely all the pixels originated were "good pixels" containing information) it wouldn't look better if it's transported in 2K or 4K.

Let me explain that another way so that it can be more clear. Pretend we shoot a film "4K". It would look just as good to have it resized to 2K in a lab and sent to all theaters as 2K and be played back at 2K than to send it as a 4K package and have the equipment resize it and display it 2K.

In that sense, 4K-converted-to-2K does not look any better than 2K.

What's happening is that most 2K films are actually either HD originated (i.e. 1920x1080 16:9 4:2:2 10bits) or DI'ed (i.e. 2048 x 1556 4:3 4:4:4 12bits).

Now, due to aliasing and nyquist law, in order to extract true 2K detail from something w/o an aliasing frequency cutoff filter you must scan at least at 4K (i.e. supersample it).

That's were a lot of the "confusion" comes from when people think of film as containing 4K detail. 4K is the recommended fixed-grid pixel count to scan a Super35 full frame in film so that the "true" 2K full-color detail can be obtained, with good edge detail and with no aliasing. It doesn't necessarily mean that film, even a first generation negative, can resolve true 4K details.

I guess what I try to say is that most movies originated as "2K" probably contain more like 1.6K in detail while most movies originated as "4K" actually contain something like 2.7K-3K in detail.

Thus, when you digitally resize the 4K one (say true-2.7K equivalent) to 2K you actually get "true" 2K detail and very little/no aliasing.

But when you originate a movie in 2K you usually only got say 1.7K in detail plus a little aliasing and that's what you are displaying at best in 2K, of course.

Most of this is due to current limits on camera/scanners and physics Soon, in a few months, more of the better 4K cameras will hit the market (i.e. 6K-stripes or 9K-bayer cameras capable of almost-true 4K resolutions without aliasing, even RED will get an upgrade to 4.5K-bayer). People using those cameras are going to output a 4K transport that's close or actually "true-4K", edit it 4K, and would want it to be displayed in 4K if it was possible.

Meanwhile I don't see a whole lot of improvements in the 2K camp as it has been widely superseeded by >4K equipment, even in small-camera factors etc.

This is a transitional period where the 2K cameras are dying and the new (and improved) 4K cameras are coming in. If it weren't because of DLP manufacturing/price restriction, this would also be happening in the exhibition front. But no DLP 4K projectors are foreseable in the near future, so no hurry from producers to do 4K either in the short run. Alternative 4K projection methods are not all there for Dcinema applications just yet, in spite of Sony's brave offerings.
 

---