[rlsmith]

The Society for Information Display is having its annual convention in San Jose this week. I went to the exhibits; here is my review.


Most of the stuff is not very relevant to HT. A lot of component makers, a lot of LCD computer screens. Here is the HT stuff:


1. Phillips had their LCoS on display, in a large RPTV format WXGA running 720P from ABC. It looked very very good. The image was very solid, and the color range quite good. For once, I could see some snappy pastels on a digital picture.


Phillips--which had the biggest booth--also had some UXGA LCD screens, some prototypes. Very nice.


2. NEC was showing some very nice plasmas. The claim a 700:1 contrast ratio.


3. Sony had an apparently new kind of LCD. Very flat screens. They were demoing SVGA.


4. JVC was there with a D-ILA in an RPTV box. Looked very good.


5. Sage/Faroudja had their "flag" demo of their interlace/3:2/DCDi clip. Very impressive. The demo had two LCD's side by side, each taking 480P from a different DVD player. The first was a Toshiba progressive scan machine using the Genesis chip. The second was a not-yet-released Kenwood player with the Sage chip.


The flag is waving in the breeze, and you get to see the interlace and aliasing artifacts and how they are handled. Incredible difference! This is an impressive demo.


I spoke quite a while to the guy. He says that a lot of manufacturers are lining up to use their chip. I guess its good news that Kenwood--generally a budget manufacturer--is one of those (rather like the Skywood player that has received a lot of attention on this forum).


I asked him about the new Infocus 530 with the Sage chip. He said he has seen it and it is very impressive. I asked about the color wheel and he winced a bit, saying, yes, it does have the dreaded clear section, but that it looks very good.


[I should mention that Silicon Image and Genesis were also there but I didn't spend any time at their booths, having found a better toy at Sage.]


Anyway, that was 90 minutes at the show.


Let me know if you have any questions about what else was there and I will try to answer them.



Bob

 

[John Mason]

Thanks for the report, Bob. Curious if you noted any other feature being promoted for the Sage/Faroudja chip besides enhanced deinterlacing. Know it's hard to catch much from an exhibit booth, but thought they might also be stressing some special patented color-enhancing features carried over from the high-priced Faroudja units.


(Last year I waded through a very lengthy abstract summary for all the papers presented the previous year. Quite remarkable, all the variations in new technology. It's available online at sid.org. And believe you can join SID to obtain the full papers.) -- John


------------------

STOP DVI/HDCP AND DFAST

 

[rlsmith]

I know that they have DCDi. I did not hear or read anything about color enhancement.


The two displays looked identical WRT color. But they were just LCD screens that might not reveal much anyway.


The rep told me that Sage is going to be very aggressive about getting the Faroudja technology out there. Hopefully, this means pricing that people other than tv networks and .COM millionaires can afford.


He also said that they tried to find a DVD player with built-in Silicon Image chips for another comparison point. The only machine that they could find had been recalled. So the only demo they had was Genesis.


I think that a three-way shootout with Silicon Image would also be helpful.


But I was very encouraged by the sign that Sage intends to actually get the Faroudja technology out there.

 

[rolypoly]

re:sage faroudja technology. I also had a chance to speak a little with the Sage rep. They stated the Kenwood model should be released Oct. of this year for $1200. However, they also said that a few months later, we would see several more units in the $300-400 range that would have the faroudja de-interlacing.


The $1200 Kenwood unit also did have the color enhancing technology which is supposed to bring out sharper lines and edges; however, it was not enabled, and I did not get much more information from the rep.

 

[johnD]

I'm new to the forum and noticed some comments on the Philips LCOS system that was shown at SID and INFOCOMM. Since I helped develop the system, I thought I'd offer some technical info (much of this and more can be found in Jeff Shimizu's paper presented at SID):


Under the brand name of "engaze", Philips showed two research rear projection prototypes at SID: a 64 inch and a 36 inch. Only the 36 inch was shown at Infocomm. Material included infocomm slides and 1280x720 HDTV. Both used the same light engine which consisted of a single 1280 x 768 panel (only 720 rows were addressed), 180 Hz frame rate, and scrolling prisms to generate phased color stripes. Everything was developed by Philips including the panel.


Since the light engine uses only 1 panel there are no convergence problems. Therefore the perceived resolution is better than a 3-panel 1280x720 system. The 1-panel approach also lends itself to a lower cost light engine because of the lower tolerances on optical components. And of course it never goes out of convergence.


The gray scale rendition is better than any other 1-panel system because the panel is able to display true 8-bit logarthimic grays at each pixel location. This is because although the signal remains digital from source to panel, it is converted to an analog voltage within the panel at the pixel. Therefore the system does not suffer from the contouring problems common in DLP, FLC, or PDP systems. (If you don't see contouring in these systems it's because the designers add noise (error diffusion) to the video to hide the inability to show 8-bits per color component. So instead you see shimmering noise, especially in gray scenes such as clouds, snow, dim lighted backgrounds, etc.)


The pixel turn-on and turn-off times are under 1 msec allowing each pixel to be updated at a 540 Hz rate (with an 8-bit analog value) thus giving the full color rate of 180 Hz. This effectively reduces color breakup to below the threshold that most people are sensitive to.


The panel pixel size is 20um x 20um with only a 1um gap between pixels. The resulting large aperture ratio made it very difficult to see pixels giving the picture a very smooth, quiet appearance.


The result is a system in which you do not see the screen door effect and you do not see any noise at all other than source noise, including the dark scenes. Most visitors were amazed by the quietness, richness, and quality of the images. The initial target market is consumer HDTV.


John Dean

 

[arrow]

John,


Is the same LCOS system used in the Hitachi shown at Infocomm or is that completely different technology?


--Les

 

[RicardoD]

John,


Are their any plans to bring Philips LCOS engine to front projection devices?


Ricardo

 

[johnD]

I believe Hitachi showed a 3 panel front projector -- totally different technology. As far as Philips development plans, I can say that Philips plans to sell the panels with driver ASICs as components initially while licensing the optical engine to selected customers. Light engines and rear projection sets will follow. For more see
http://www.eetimes.com/story/OEG20010601S0089 http://www.ebnonline.com/digest/story/OEG20010604S0093 http://biz.yahoo.com/prnews/010604/sfm123.html


(The eetimes article has a typo -- the panels are 1280 x 768)


--John

 

[bxa]

According to TI, DLP has effective resolution of 14 bits per pixel, it too controls ligth by rapid on/off movement. Contouring effects are not due to inability of the chip to reproduce grayscale, but rather because of de-gamma process using 7 bits in the mapping table.

There are DLP projectors out there with 10 and more bits of color resolution (per color), see vistagraph.

 

[johnD]

I admit I'm not a DLP expert, but here goes:


First of all the DLP produces light with a linear brightness characteristic which means they need at least 10-bits per pixel (and probably more for the blacks) to produce the equivalent of 8-bit logarthimic brightness display which is the minimum necessary for a non-contouring display. This 10-bits must be applied to each color component at each pixel (for a total of 30-bits/color) if you want to achieve "true color" without contouring. From everything I've seen and heard, the TI mirrors can't switch fast enough to show 30-bits per pixel (with digital PWM light systems, you are limited by your ability to switch LSBs fast enough) on a single panel at the frame rates required. Therefore they must apply error diffusion (spatial dithering) to the video to mask the contouring. That's why all the single panel DLP systems I've seen look noisy.


For this and other reasons, 3 panels are used in the high end DLP systems and 1 panel in the low end DLP systems. With 3 panels you have at least a 6x factor extra in time to load the panel with video (3x factor because of 1 color component per panel and 2x or so factor because frame rates are lower since there is no color break-up). I suppose with that much extra time you could do 10 or so bits per color component but I'm not sure what the 14-bit number means. Could it mean 14-bits per pixel with red 5, green 5 and blue 4?


-John

 

[noah katz]

John,


Welcome to the forum.


What contrast are the LCOS capable of?


Thanks


------------------

Noah

 

[Robert Whitehead]

This is not entirely germaine to this forum, but worthy of noting. The Sage rep said Sage "tried to find a DVD player with Silicon Image chips for comparison [but] the only machine they could find had been recalled."


This is simply not true. There are four available players with the SI/DVDO503 deinterlacing chip: Ayre, Camelot Roundtable, EAD Theatervision-P, and Denon DVD-2800. NONE of these has been recalled.

 

[rlsmith]

Robert Whitehead,


I talked to the Sage rep at the SID as well. He told me something slightly different, namely that the DVD player that they had ARRANGED to get with the Silicon Image chips turned out to be recalled; he allowed that there were others on the market with the SI chips but that he hadn't arranged to get them. I specifically mentioned the Camelot, and he said that he had not compared the Camelot to the Sage.


The Toshiba that was being used at the Sage demo (the flag demo) in the comparison had the Genesis chip set.


While attending Infocomm, over at Infocus, which is now using the Sage chip set in their 530, the product manager told me that they had evaled the Sage, Genesis and SI sets before deciding on the Sage. Infocus previously used the Genesis in their 340/350. (The new 340B has some really inferior chip set they are saving money on, and should not be considered for HT.)

 

[johnD]

Noah,


We measured 450:1 full white to full black system contrast in a dark room prior to SID for the 36 inch projector.


John

 

[Robert Whitehead]

What the Sage rep told you does not really alter what I said. There are four SI/DVDO505 based DVD players available and NONE of them has been recalled. Whichever one Sage had supposedly intended to use and claimed was recalled was, in fact, NOT recalled.


In the early Skyworth posts in the DVD forum Stacey and/or Don reported that they had compared the FLI2200 and SI/DVDO503 and reported that they found them virtually indistinguishable except for a few instances on video material where the Sage pulled ahead, but you had to look closely to notice these. Perhaps this is the reason the SI based player was "recalled."

 

[KFung]

The TI Cinema projector does 14-bits/colour, of course that's with 24 frames/sec source material. As far as I know the limitation on switching speed for the DMD device is limited by the speed of the driving electronics, not the mechanical limitations of the device itself.


Regards,


Kam Fung

 

[johnD]

Let's assume you want to show 10-bit color video on a 1-panel DLP system at 240 Hz. If you address the DLP in a straight-forward pulse width modulated fashion, The LSB time is (1/240)/(1024*3) = 1.36 us. A mirror switches in about 5 usec according to white paper by Meier at: http://www.dlp.com/dlp/resources/ove...asp#challenges


So if you look at the mirror speed alone, you're limited to 8-bits per color best case. As mentioned, this is not enough to produce a natural image due to the linear signal to light transfer characteristic. If they are also limited by the input bandwidth to the chip, then things only get worse. The remedy for this is error diffusion, see: http://www.dlp.com/dlp/resources/whi.../pdf/vproc.pdf


For 3-panel at 24 Hz and 14-bits, the LSB time is (1/24)/16384 = 2.54 usec. Again, not possible due to mirror speed. However 13-bits is possible according to this calculation.


Of course they may steal back some of the time with processing tricks such as assigning different bit weights to each color component.



-John

 

[noah katz]

John,


"We measured 450:1 full white to full black system contrast in a dark room prior to SID for the 36 inch projector"


Quite respectable for a measured value; thanks.


------------------

Noah

 

[CINERAMAX]

excuse me RL3


" Sony had an apparently new kind of LCD. Very flat screens. They were demoing SVGA."


panels or projectors what size?

 

[DaViD Boulet]

John,


This sounds VERY promising. Any idea on a time-frame for this technology to enter the Front-projection market?


BTW...anybody know what's up with that new technology Sony had been working on?