AVS Forum banner
Status
Not open for further replies.
1 - 20 of 20 Posts

·
Registered
Joined
·
1,730 Posts
Discussion Starter · #1 ·
I wanted to ask about this a while ago when I noticed it is last months WSR..


Here is a quote from Gary Reber's editorial this month:

Quote:
As a repeat (see Issue 66, November), the following is what industry video guru Joe Kane (creator of Video Essentials and Digital Video Essentials) perceives as the differences between CRT-based display quality and DLP or fixed pixel. Real blacks are one aspect of the CRT, which is a difficult parameter to obtain in a light bulb-based display device. “It is hard to shut it off in a very small area of a picture, let alone a whole picture,†said Joe. “Then, too, part of the reason I like CRT is it is analog. When it comes to vision, all images in the digital domain have to be converted to analog at one point or another in the output path or we won’t be able to see it, just as all the audio has to be converted back to analog or we don’t hear it. Even in a DLP projector, where the conversion takes place at the transition, and we are delivering a digital signal to the mirrors and a digital signal is moving those mirrors, the moving of those mirrors with the light comes back to an analog image,†he explained. “What I end up seeing in the CRT analog display device is a continuous display, with a smooth transition from one pixel to another. If one pixel is low in level, and the adjacent pixel is high in level, in the analog domain we can create a transition between those two pixels—a ramp that goes from dark to light.


“With DLP, when we talk about 1920 x 1080 or about 1280 x 720 pixel sizes, we have an exact pixel match for the source. The reality is, if we want an analog-looking picture, we have to have 2x the number of pixels in each direction before we can create a transition from one pixel to another that looks like analog. And those pixels have to not have an unused area around each of them, a maximum amount of the pixel area has to be used,†said Joe. “I am not telling you that fixed pixel displays do not have their place. CRT displays have such huge limitations, both in maintenance and in light output. Still, I prefer CRT. I can get the best looking image out of CRT, but I have to work hard to get that image out of it. Fixed pixel displays will get my attention when we can do 4000 x 3000, when the pixel array is very much larger than the source. At that point, we can begin to get what looks like an analog picture out of those display devices.â€
I take this to mean that fixed panel display technologies (LCOS, D-ILA, DLP, LCD, Plasma, GLV) will need to output better than 1080p to do HD proper justice...


Can someone explain to me why a 1080p D-ILA or DLP wouldn't be enough?
 

·
Registered
Joined
·
5,574 Posts
This sounds like the standard "analog is better" argument.


The argument essentially assumes that analog is capable of "infinite" resolution whereas digital is fixed. The problem with this argument is that analog isn't infinite at all but has its own built-in inadequacies, which may be a bit harder to pinpoint exactly.


If you put your face up to an analog picture, you will see defects and artifacts of the structure of the image. Take the best CRT you can name (is it the Sony G90 for example), you wil see lines etc. if you get closely enough.


Surely more pixels might in principle be better but I am not sure that you need 4000 pixels to be better than a CRT. The JVC QXGA projector beats any CRT or digital image I have ever seen, bar none. It is 1536X2048. My guess is that a true 1080X1920, with small gaps like LCOS, will consistenty beat any CRT system for resolution and clarity. Blacks and contrast may or may not still have a distance to go.


BTW film is still better IMHO, especially 70mm.


We need to stop romanticizing these things and get real about the overall issues that effect quality.
 

·
Registered
Joined
·
4,351 Posts
Completely sloppy reasoning from someone who is supposed to be highly influential in the business.


For example, CRTs are only "continuous" in one direction, horizontal. Or has Joe Kane forgotten about the existance of scan lines?


Andy K.
 

·
Registered
Joined
·
6,857 Posts
I've actually thought about this very logic.


On the one hand he's right...for it to look "analog" in the sense that he's describing, yes, a smooth transition is required.


However, there are a few perspectives that he's not taking into account:


1. the "ramp up" principle, as he calls it, with a CRT is actually a FLAW of that projection technology. If the source image had a severe and instantaneous transition...from dark to light for instance...the CRT is *flawed* because it may have difficulty reproducing this abrupt change.


2. The relevance of any of this is entirely related to viewing distance. Given a 1.5 viewing distance (width to distance ratio), it's quite possible that 1920 x 1080 pixels may contain enough inherent information that they can reproduce a visually smooth image without any further oversampling (really what he ought to be describing). In nature there are subtle and gradual changes from dark to light and the resolution of 1080P turns out to be sufficient at that viewing distance, then as long as those gradations are captured in the 1920 x 1080 image then they should be rendered visually in tact on the screen.


However, in nature there are also instantaneous and severe transitions like a black boweling ball in front of a white wall. Any display technology that "smooths" the transition from the edge of the ball to the wall because of its inherent design is FLAWED just like any device that would artifically sharpen a smooth transition (LCD) is also flawed.


IMO, most CRT displays blur and sacrifice some level of detail to render the "smoothing" affect of the crt gun. It's like suggesting that speakers have tweeters that add coloration to the sound to compensate for bright recordings. There are both mellow and bright sounds in recordings. Speakers should render them faithfully. As long as no artifical "sharpening artifact" is present from a given viewing angle then a digital projector's resolution is sufficient for the task.


In any case it's very simple to determine if you need more resolution than 1920 x 1080. Do you see pixelization artifacting from your preferred seating distance? If you don't...then no additional pixels are required to "smooth" the image.


That being said, I'll WILL aggree that doubling the pixel count (or higher), with proper scaling, should result in an even *better* image. But I'll bet that a good LCOS-based projector with 1920 x 1080 produces an image equal or better to almost any CRT in terms of smoothness and detail, and I'd wager that a 4000 x 3000 res digital projector would look VASTLY better than any CRT machine...not simply as good as.


Also, as another poster thoughtfully mentioned,


3. CRT displays are analog in nature in their horizontal domain, but are quantized and therefore digital in their vertical domain (scan lines).


Sorry Gary!
 

·
Premium Member
Joined
·
4,827 Posts
Not necessarily to support Joe's theory, but just to clarify one point....


Scan lines on a CRT face can be taken out of the equation through optimal scaling--assuming that the projector has sufficient bandwidth to handle said scaling--taking into account the spot beam size and the available raster area. That's an "advantage" of analog projector--it is in effect infinitely adjustable. Even if the optimal scaling setup cannot be achieved, as someone above noted, if you can't see the scanlines from a desired viewing distance, their impact disappears. If I press my face to the screen, I can see (barely!) scanlines on 1080p material on my projector, but at 4' I can't see them at all. This is why triplers, quadruplers and scalers have been tied to CRT front projectors since the old days--to minimize or "disappear" (watch too much Sopranos) the scan lines on the screen.


That said, I'm not trying to start a war...I eagerly await the release of a long throw digital projector that can meet or beat my current setup--bring it on!
 

·
Premium Member
Joined
·
25,501 Posts
"The reality is, if we want an analog-looking picture, we have to have 2x the number of pixels in each direction before we can create a transition from one pixel to another that looks like analog."


You could do the same thing with a slight defocusing to blur/blend the the pixel edges, at a much lower cost for the display and video processing.
 

·
Registered
Joined
·
2,066 Posts
Joe Kane advocacy of 720p against 1080i, which will inevitably lead to 1080p anyway, destroyed all credibility to his advices as far as I am concerned.


I had the occasion to compare 720p to 1080i properly de-interlaced by the JVC QX1, which showed a clear supremacy of 1080i. The fact that Joe advocates 720p is a clear indication that he does not know what a fixed 1080p panel can do.
 

·
Registered
Joined
·
4,351 Posts
Alex, the fact that you can, ideally, exactly space scan lines beside each other so that no gap appears is irrelevant. The same can be said of any digital projector - if you sit far enough away.


In the case of DILA, you dont even have to get 4' away before you stop seeing the gaps inbetween pixels.
 

·
Premium Member
Joined
·
4,827 Posts
kromkamp, I think you're missing the point: an analog projector is the ONLY current display type that can approach "no gaps." I was simply noting that on *my* CRT front projector (G90) that *I* can see scan lines at 1080p/72...all that says is that I have more room to scale the signal to make said scan lines disappear. I choose to output DVD at 1440x960p/72 not to minimize scan lines (although this gets pretty close), but because it's an easy scaling multiple for film-based 720x480p/24 sources...I could easily bump up to something that would visually obliterate scan lines even at point blank range (say in the 1200p range), but to do so would increase the incidence of scaling artifacts in *my* setup. You made a comment that scan lines were a limitation/fault of analog projectors; I simply countered that they can effectively be removed from the equation. This post is simply to restate that it can be done at ANY viewing distance, even 1", *if* you wanted to do it.
 

·
Registered
Joined
·
4,351 Posts
Well, I dont think I am. Sure, its a nice curiousity to know that you can remove the scan lines even at 1" away, but:


1)Its meaningless because, for all intents and purposes, you can remove the "scan lines" (meaning gaps in between pixels) for at least 2 digital technologies (DILA and DLP) for all useful seating positions. Both technologies very nicely approximate a continuous image at all but the most extreme (and uncomfortable) viewing distances.


2)The very fact that you have scan lines means a fixed, discrete number of pixels in the vertical direction, whatever that number may be. Since Joe Kane is expounding on the "continuous display" aspect, it is worth pointing out that its simply not true. There is a real, measurable vertical resolution for CRT projectors.


Andy K.
 

·
Registered
Joined
·
1,948 Posts
any crt device that I know of has triads of three color phosphors (or an equivalent) that act like discrete, dare I say, "digital," pixels. the notion of analog "continuity" of either scan line or scan line content is limited to the dot pitch. Now this may be a very fine pitch compared to real pixels in a digital display, but the point is, the analogy of crt to other analog audio devices ultimately isn't accurate.
 

·
Premium Member
Joined
·
4,827 Posts
Andy, even at an industry best 93% fill, even D-ILA can't "rid" itself of black space surround the fixed pixel elements. I can think of a number of industrial/imaging applications where even a 93% fill won't cut the mustard, thus the continued--but uber niche--application of CRT front projectors to do said jobs. Again, not trying to throw gas on a fire, but it seems that some people in the conversation don't/didn't understand that *visible* scan lines (at ANY viewing distance) are simply an artifact of not driving the CRT projector as its optimal settings (each CRT front projector has one sweet spot)...kind of like only lighting up 360 rows in a 1280x720p fixed rate display. Not trying to say anything more, nor anything less. I'll let Joe Kane defend himself, thank you very much! :)


John, Joe Kane was refering to CRT projectors (specifically, CRT front projectors), which use individual CRT tubes to create the three primaries. The limitation you speak of is for direct view CRT monitors, which use a single gun.
 

·
Registered
Joined
·
1,948 Posts
isn't each color tube still a collection of discrete phosphors? or am I mis-understanding it?
 

·
Registered
Joined
·
1,460 Posts
jlm,

Each color tube has a face with continuous phosphor.


This phosphor is illuminated (excited) with an electron beam that scans horizontally in vertical steps.


This is why CRT's are horizontally "continous" and vertically "discontinous" depending upon the vertical scan rate.


If the vertical scan rate is high enough to have the spot in the first vertical step touch the spot in the next vertical step, then you will see no scanlines - thus the sweet spot in vertical rez discussed above.


[All this and I own digital :)]
 

·
Premium Member
Joined
·
4,827 Posts
John, yes, each tube is coated with a specific class of phosphors, but I couldn't even begin to estimate what the diameter of said material would be in a typical CRT front projection application (phosphors are typically measured in the tenths to very low hundreds of ìm, I think). I do know, though, that phosphor size can be altered to suit OEM needs. Phosphors are also layered during application, so I don't think it's like there's a matrix lineup of A x B individual discrete phosphors on the tube face.


If you want to hurt your brain, the answer you seek may be in this publication, but I can't tell. Yikes.

http://www.cl.cam.ac.uk/~mgk25/ieee02-optical.pdf
 

·
Registered
Joined
·
43 Posts
Quote:
Originally posted by robena
Joe Kane advocacy of 720p against 1080i, which will inevitably lead to 1080p anyway, destroyed all credibility to his advices as far as I am concerned.


I had the occasion to compare 720p to 1080i properly de-interlaced by the JVC QX1, which showed a clear supremacy of 1080i. The fact that Joe advocates 720p is a clear indication that he does not know what a fixed 1080p panel can do.
Mathematically 1080i is the same as 540p (ignoring horizontal resolution).

So Joe is right in saying that 720p is better than 1080i. You are comparing

720p with 1080p and of course 1080p wins. But Joe is still right.


One area where he is wrong: the eye is digital (it has 120 megapixels).

So video doesn't have to be converted to analogue at some stage.


I don't see how a 1500 line CRT is the same as a 3000 line digital. Maybe

he is thinking that the spot is not of uniform intensity and thereby actually

has 3000 line resolution. Sounds dodgy. I disagree with him there. But

certainly horizontally the analogue spot will emulate a higher sampling

frequency. The question is whether or not that emulation conveys any

extra information to the brain. It might but if you can't see the pixels on

the digital display then it won't make any difference.
 

·
Registered
Joined
·
6,857 Posts
Quote:
Mathematically 1080i is the same as 540p (ignoring horizontal resolution).

So Joe is right in saying that 720p is better than 1080i.
Sort of. the scanning rate is the same, but a 1080I signal of a static image (without motion) does indeed capture more visible vertical information than a 540P image. In fact, interlacing was the conceived of to provide *twice* the vertical resolution with the same scanning frequency...that's why we have it with NTSC in the first place.


Where interlaced signals have no more resolution than the scanning-rate-equivilent progressive counterparts is with fast motion (because only one field at a time is being used to caputre and record the detail for a given image). And of course there are issues with aliasing and related artifacts, and so vertical filtering is applied in common practice which *does* effectively cut your vertical resolution significantly. However, it's important to realize that with still images or images lacking in fast motion, 1080I *does* contain more resolution than 540P.

Quote:
The shadow mask in a CRT also makes it a 'neo-digital' projector. - By the way, I don't know enough about the projector CRT to offer my opinion, and I could be way off, though.
Front projection and rear-projection CRT guns have no shadowmask. The shadowmask is something that direct-view CRT displays possess which, in part, allows them to display all 3 colors at once on the same display surface. BTW, in that regard, direct-view CRT displays do have a visible "pixel" structure that I find distracting in all but the very best designs.
 
1 - 20 of 20 Posts
Status
Not open for further replies.
Top