[Nielo TM]

A 47” 1080p Philips LCD with LED backlighting will be releases this year according to engadget. So if you have some large amount of cash to burn, here’s your chance.







PS: I don’t see the point of releasing SED, FED or the next-gen PDP anymore lol (unless they can be manufactured cheaply)

[Tejstar]

Interesting - it looks like that Sharp set has some competition! Would love to see one in the flesh!

[NicolasB]

Originally Posted by Nielo TM

The ultra high contrast ratio is achieved by having dedicated LED backlighting for each individual pixel.

I don't see any reference to 1 LED per pixel in the article you link to - do you have a source for that? If it really does have 1 LED per pixel that would be an amazing advance.

[DanDT]

Originally Posted by NicolasB

I don't see any reference to 1 LED per pixel in the article you link to - do you have a source for that? If it really does have 1 LED per pixel that would be an amazing advance.

This set doesn't have 1 LED for each pixel, that would be totally overkill and a waste of resources at the moment. Plus, i don't even think it's possible to release a set this year like that - the technology is just not there. It most likely has a array of LEDs at much lower resolution than the real LCD's resolution, which is really more than enough and will give much, MUCH better results than normal LCDs anyway.

[Dave163]

Can I ask a really stupid question? Rather than having an array of hundreds of switchable LEDs behind the LCD screen instead of a conventional backlight, why not have a conventional backlight but mount a second LCD screen between the backlight and the main LCD screen, immediately behind the main LCD screen (ie the one with colour pixels)? The secondary screen would need only black-and-white pixels capable of a greyscale, and they could be much larger than the tiny colour pixels. The purpose of the secondary screen would be to adjust the amount of backlighting at localised areas of the screen to improve the contrast ratio, ie letting all the light through in light areas and blocking off most of the light in dark areas. I'd have thought that would give pretty much the same result as having an array of dimmable white LEDs, but at much reduced cost. Can I patent the idea???
Dave

[NicolasB]

Originally Posted by Dave163

Can I ask a really stupid question? Rather than having an array of hundreds of switchable LEDs behind the LCD screen instead of a conventional backlight, why not have a conventional backlight but mount a second LCD screen between the backlight and the main LCD screen, immediately behind the main LCD screen (ie the one with colour pixels)? The secondary screen would need only black-and-white pixels capable of a greyscale, and they could be much larger than the tiny colour pixels. The purpose of the secondary screen would be to adjust the amount of backlighting at localised areas of the screen to improve the contrast ratio, ie letting all the light through in light areas and blocking off most of the light in dark areas. I'd have thought that would give pretty much the same result as having an array of dimmable white LEDs, but at much reduced cost. Can I patent the idea???
Dave

I think part of the reason for using LED backlights is that they're not white - there are separate red, green and blue ones. This is (I believe) what allows LED-backlight devices to achieve a much wider colour gamut than conventional LCDs: the standard backlight tends not to be very well balanced in terms of red, green and blue components, and it's trickier to get the filters the right colour.

If all of the light hitting the LCD panel is already within the correct R, G and B bands, that makes life much simpler. It also means you can (for example) dial down the blue component of the backlight while leaving the red and green components at full brightness, so you have rather more precise control over the output.

In your proposed setup you'd also have to ensure that the light coming through the back grid had the correct spectral characteristics (e.g. the same balance of red and blue) at all brightness levels, which would be quite a trick. And you'd have alignment problems: how do you ensure that all of the light from one cell in the rear grid only strikes the appropriate front pixels and nothing around them? The only way to do this would be to make the lightly highly collimated, which would reduce your viewing angle almost to zero.

It'd also be expensive, of course.

I think a more interesting question is: is anyone currently developing an LCOS device using LED lighting? That would kick *rse.

[Nielo TM]

I don’t see how that level of contrast ratio can be achieved without having very deep blacks.

[Dave163]

Originally Posted by NicolasB

I think a more interesting question is: is anyone currently developing an LCOS device using LED lighting? That would kick *rse.

Thanks Nicolas!

To answer your question, LG are taking a twin-track approach to projection technology: on the one hand they have a LCOS rear-projection TV:
http://www.cinenow.com/uk/view-press-release-312.html

On the other hand they have an LED-lit DLP projector:
http://www.aboutprojectors.com/news/...led-projector/

No mention on the lamp source for the RPTV, so presumably it's a conventional bulb. But if they stuck a white LED source into it, it would be really interesting to compare that to a DLP.

Hard to know which would be better - I can't stand DLPs due to rainbows, but if the new Samsung RPTVs have fast enough colour switching to exceed what the eye can perceive, you have the plus that there's only one microarray so no risk of misconvergence. On the other hand, with LCOS you have no screendoor, but risk of misconvergence of the three colour panels. I'd heard that DLP has better contrast than LCOS though I'm really impressed by the new Sony LCOS (can't afford one though!) All said, I prefer the "look" of DLP to LCD but can't stand the rainbows.

Thoughts?

[NicolasB]

Originally Posted by Dave163

No mention on the lamp source for the RPTV, so presumably it's a conventional bulb. But if they stuck a white LED source into it, it would be really interesting to compare that to a DLP.

Assuming it works the same way as existing LCOS devices, i.e. that it uses three separate LCOS panels, it would seem to be a particularly silly idea to use white LEDs. It would make more sense to use red LEDs for the red panel, green for green, and blue for blue.

Originally Posted by Dave163

Hard to know which would be better - I can't stand DLPs due to rainbows, but if the new Samsung RPTVs have fast enough colour switching to exceed what the eye can perceive, you have the plus that there's only one microarray so no risk of misconvergence.

I remain suspicious about the alleged rainbow-free nature of LED-powered DLP devices. You're still reducing the problem rather than eliminating it: the eye will move less between successive sub-frames, but it will still move, and the inages still won't precisely line up on the retina. There's also the question of how fast the actual DLP mirrors can switch.

Originally Posted by Dave163

On the other hand, with LCOS you have no screendoor, but risk of misconvergence of the three colour panels. I'd heard that DLP has better contrast than LCOS though I'm really impressed by the new Sony LCOS (can't afford one though!) All said, I prefer the "look" of DLP to LCD but can't stand the rainbows.

DLP traditionally has very good ANSI contrast ratio, but I believe that in terms of absolute black level and shadow detail (i.e. differing levels of shadow in an overall-dark scene) LCOS has superior contrast to DLP; doubly so in a dynamic-iris system. (I'm actually not sure how LCOS performs on ANSI contrast tests; most oif the "DLP has the best ANSI contrast" claims date back to before LCOS became a serious contender).

[sandstheman]

Originally Posted by Nielo TM

I don’t see how that level of contrast ratio can be achieved without having very deep blacks.

The way they achieve it is by switching off clusters of LEDS in pure black areas. The problem with current backlight setups is that the backlight is constantly on and they depend on the LCD being able to shut out most of the light.

Although i would take the figures with a pinch of salt, as you can have a display that outputs 0 cd/m2 at it's darkest and 1 cd/m2 at it's brightest, not very bright but by the virtue that it is outputting 0 for black it technically has an infinite contrast range.

For a good explanation have a read of this article, it's pretty much an identical principle:

http://www.bit-tech.net/hardware/200...hdr_edr/1.html

[Nielo TM]

Thx for site, it explains a lot. Anyway, I am aware of manufactures overstating the contrast ratio of the display by increasing the brightness, contrast and backlight to maximum (e.g. LCD 600cd/m2 and PDP 1200cd/m2) and measuring the white level, then lowering the contrast, brightness and backlight to measure the black level (LCD 0.3cd/m2 and PDP 0.2cd/m2). By doing so, they will achieve very high contrast ratio (LCD, 600 divided by .03 = 2000:1 and 1200 divided by 0.2 = 6000:1 for PDP).

I'm just gonna wait for OLED lol

PS: Hope the warranty covers single dead LED, unlike the pixel policy lol

[DanDT]

Originally Posted by Nielo TM

Thx for site, it explains a lot. Anyway, I am aware of manufactures overstating the contrast ratio of the display by increasing the brightness, contrast and backlight to maximum (e.g. LCD 600cd/m2 and PDP 1200cd/m2) and measuring the white level, then lowering the contrast, brightness and backlight to measure the black level (LCD 0.3cd/m2 and PDP 0.2cd/m2). By doing so, they will achieve very high contrast ratio (LCD, 600 divided by .03 = 2000:1 and 1200 divided by 0.2 = 6000:1 for PDP).

I'm just gonna wait for OLED lol

PS: Hope the warranty covers single dead LED, unlike the pixel policy lol

On LED LCDs, there would be no point because when showing the same picture at high contrast, some of the LEDs will be completely OFF, giving pure black, and some LEDs in another area of the screen will be ON, giving full brightness. And the areas in between will have LEDs which are dimmed.

These sets will give picture quality that is SO above current LCDs that it will hardly matter what the real constrast ratio is. I do agree that this new trend of 1,000,000:1 CR's is pure marketing. Mathematically, the sets would have "indefinite" contrast ratio.

[sandstheman]

Originally Posted by NicolasB

I think part of the reason for using LED backlights is that they're not white - there are separate red, green and blue ones. This is (I believe) what allows LED-backlight devices to achieve a much wider colour gamut than conventional LCDs: the standard backlight tends not to be very well balanced in terms of red, green and blue components, and it's trickier to get the filters the right colour.

If all of the light hitting the LCD panel is already within the correct R, G and B bands, that makes life much simpler. It also means you can (for example) dial down the blue component of the backlight while leaving the red and green components at full brightness, so you have rather more precise control over the output.

In your proposed setup you'd also have to ensure that the light coming through the back grid had the correct spectral characteristics (e.g. the same balance of red and blue) at all brightness levels, which would be quite a trick. And you'd have alignment problems: how do you ensure that all of the light from one cell in the rear grid only strikes the appropriate front pixels and nothing around them? The only way to do this would be to make the lightly highly collimated, which would reduce your viewing angle almost to zero.

It'd also be expensive, of course.

I think a more interesting question is: is anyone currently developing an LCOS device using LED lighting? That would kick *rse.

I don't think these panels will be using arrays of RGB led's as white LED's themselves have a much better colour GAMUT than the flourescent backlights that are currently used, but also the problem of getting LED's small enough and powerful enough and running cool enough to have 3 per pixel has not really been solved yet. I would say in all the cases of the new LED LCD TV's they will be using arrays of White LEDs (which are actually Blue LED's with a phosphor coating to create white: http://dansdata.blogsome.com/2006/12...-diffraction/). Again if you have a read of the article i linked in an earlier post you'll see an outline of how this is supposed to work and the related heat issues

[NicolasB]

Originally Posted by sandstheman

I don't think these panels will be using arrays of RGB led's as white LED's themselves have a much better colour GAMUT than the flourescent backlights that are currently used, but also the problem of getting LED's small enough and powerful enough and running cool enough to have 3 per pixel has not really been solved yet.

There's no way of having an individual white LED for each pixel either. Your previously-linked article, for example, describes an array of 45x31 LEDs driving am entire hi-def screen. (This, incidentally, means that the ANSI contrast ratio would be no better than that of current LCD displays - only the on/off value would be improved).

Originally Posted by sandstheman

I would say in all the cases of the new LED LCD TV's they will be using arrays of White LEDs

You may well be right. I imagine DLP displays will use three colours of LEDs, though, lit in sequence (for a single-chip device) or lighting three separate panels (for a 3-chip device). That would also be the logical arrangement for an LED-driven LCOS display.

[sandstheman]

Originally Posted by NicolasB

There's no way of having an individual white LED for each pixel either. Your previously-linked article, for example, describes an array of 45x31 LEDs driving am entire hi-def screen. (This, incidentally, means that the ANSI contrast ratio would be no better than that of current LCD displays - only the on/off value would be improved).

You may well be right. I imagine DLP displays will use three colours of LEDs, though, lit in sequence (for a single-chip device) or lighting three separate panels (for a 3-chip device). That would also be the logical arrangement for an LED-driven LCOS display.

I agree, they won't have an LED per pixel period, but i don't think they'll have as small an array as 45x31 either, LED's are small enough to have an array larger than that

[DanDT]

Originally Posted by sandstheman

I agree, they won't have an LED per pixel period, but i don't think they'll have as small an array as 45x31 either, LED's are small enough to have an array larger than that

That's not really the point. The more LEDs you have, the more power and heat the set will produce. At a certain point it becomes useless to have more LEDs. They are only needed for light source, they don't need to be very detailed, after a certain threashold.

As well as delivering real blacks and accurate colours LED backlighting dramatically improves LCD motion portrayal which is still a problem on current gen LCD despite manufacturers' misleading claims about response times. Unlike traditional always-on CCFL backlights LEDs can be turned off between frames to eliminate motion blur, simulating the impulse motion characteristic of the CRT. Philips has already demonstrated this very successfully with the strobing HCFL backlight in its ClearLCD range. It's just a pity the strobing effect doesn't work with HD resolution material.

[NicolasB]

Originally Posted by Quickbeam

As well as delivering real blacks and accurate colours LED backlighting dramatically improves LCD motion portrayal which is still a problem on current gen LCD despite manufacturers' misleading claims about response times. Unlike traditional always-on CCFL backlights LEDs can be turned off between frames to eliminate motion blur, simulating the impulse motion characteristic of the CRT. Philips has already demonstrated this very successfully with the strobing HCFL backlight in its ClearLCD range. It's just a pity the strobing effect doesn't work with HD resolution material.

Why doesn't it work with HD?

Of course scanning backlights risk reintroducing CRT-style 50Hz flicker. Maybe british TVs will be driven at 100Hz, but that's a significant step-up for the LCD panel. What about on 60Hz sources? 60Hz strobe, or 120? And what about film recovered via 3:2 pulldown? 48Hz? 72? 120? Questions, questions, questions....

[sandstheman]

I'd imagine there is plenty of scope to have the LED's strobe at varying frequencies, seeing as they don't require any 'warmup' time, which is a big advantage over normal CCFL backlights. If they have any sense they would have the TV detect the incoming signal frequency and then strobe on that basis instead of trying to conform the signal the standard the TV has been set to

[hamster]

Originally Posted by sandstheman

I'd imagine there is plenty of scope to have the LED's strobe at varying frequencies, seeing as they don't require any 'warmup' time, which is a big advantage over normal CCFL backlights. If they have any sense they would have the TV detect the incoming signal frequency and then strobe on that basis instead of trying to conform the signal the standard the TV has been set to

Yes but...you then get a beat frequency as the TV gets in and out of sync with the 50Hz room lighting...slow flicker.

I bet the thing has a Sharp panel in it....a lot of their high-end stuff used Sharp in the past.

[NicolasB]

I'd imagine there is plenty of scope to have the LED's strobe at varying frequencies

The LEDs can strobe at any frequency in isolation, but in practice they must strobe in perfect synchronisation with the LCD panel updating. The idea is that you flash each frame briefly, once per frame, at a point when the LCD matrix has finished switching from the previous frame to the current one.

Can the LCD panel switch fast enough that it can have completely finished switching from the previous frame to the current one in only 1/120th of a second? Manufacturers may quote sub-8ms switching times, but those are often calculated in a highly optimistic way. I'm not at all sure the LCD can really switch that fast. If it can't then the strobe frequency would have to drop to 60Hz which would (to many eyes) produce visible flicker.

Another problem that occurs to me (apart from the lack of improvement in ANSI contrast) is that this may introduce issues with colour- and brightness-unformity across the screen. How easy will it be to ensure that all of the LED backlights have exactly the same spectral output and exactly the same brightness? Even if they do when the set is new, will they really stay that way over the course of a couple of years, or will you start to get darker and brighter patches in different parts of the screen?

they will be using arrays of White LEDs (which are actually Blue LED's with a phosphor coating to create white)

I don't know if that's true but, if it is, that means these TVs will also suffer from screen-burn. Progress is great, isn't it?

[LV426]

Originally Posted by Dave163

...why not have...a second LCD screen between the backlight and the main LCD screen...capable of a greyscale

Would not work. LCDs work by polarising light. The light output from an LCD screen is polarised as a result (try tilting your head whilst looking at one through "polaroid" sunglasses). You can't use polarising effects on light that's already polarised (unless you scatter it first).

Originally Posted by NicolasB

Why doesn't it work with HD?

Of course scanning backlights risk reintroducing CRT-style 50Hz flicker. Maybe british TVs will be driven at 100Hz, but that's a significant step-up for the LCD panel. What about on 60Hz sources? 60Hz strobe, or 120? And what about film recovered via 3:2 pulldown? 48Hz? 72? 120? Questions, questions, questions....

Logic would seem to dictate that that each frame be shown twice to avoid flicker, but according to Philips' press release 50Hz is interpolated to 75Hz while 60Hz is frame doubled to 120Hz. I have seen ClearLCD in action and I could not detect any standards conversion judder on 50Hz SD material, which leads me to question the veracity of the press release.

As for ClearLCD strobing not working with HD signals, the only explanation I have read (which was speculative) is that ClearLCD is tied to Digital Natural Motion processing, which is not HD compatible.

Other manufacturers have been slow to offer a hardware-based solution because strobing the backlight reduces the screen brightness so much that a more powerful backlight is required; in this case, a Hot Cathode Fluorescent Lamp. Software-based 100/120 Hz black level frame interpolation (used by other manufacturers), while inferior, is cheaper and works at all resolutions.

ClearLCD does little to improve picture quality in other areas (in terms of black levels, contrast, colour saturation etc) but LED backlighting hopefully will. The ClearLCD backlight can be dimmed by 30% to improve picture quality in dark scenes but I can't say I noticed much difference from a conventional CCFL LCD.

[NicolasB]

Originally Posted by Quickbeam

Logic would seem to dictate that that each frame be shown twice to avoid flicker, but according to Philips' press release 50Hz is interpolated to 75Hz while 60Hz is frame doubled to 120Hz. I have seen ClearLCD in action and I could not detect any standards conversion judder on 50Hz SD material, which leads me to question the veracity of the press release.

Ah, but was the source material you were watching film or video? If it was film, then the conversion to 75Hz could be done perfectly: you deinterlace to reconstruct 25 progressive frames per second, and then show each one three times. But for a video source you couldn't do that; I can't think of any way of crowbarring 50Hz video into 75Hz without it looking naff. 100Hz would make more sense. (If it's capable of being driven at 120Hz, then 100Hz would not, on the face of it, seem to be a problem).

It was video material I was looking at. I agree, I don't see how you could ever get decent results from interpolating 50Hz video to 75Hz, especially on consumer equipment.

To address an earlier point you made:

Originally Posted by NicholasB

Can the LCD panel switch fast enough that it can have completely finished switching from the previous frame to the current one in only 1/120th of a second? Manufacturers may quote sub-8ms switching times, but those are often calculated in a highly optimistic way. I'm not at all sure the LCD can really switch that fast. If it can't then the strobe frequency would have to drop to 60Hz which would (to many eyes) produce visible flicker.

In the pdf I linked to above Philips presents a persuasive argument that LCD motion smearing is caused as much by objects appearing in the wrong place as by a panel's response time; killing the backlight between frames eliminates this smearing.

[Tejstar]

Some stunning pictures of a 47" set (and others) here.

[Snaques]

I may be wrong, but isn't video material interlaced and with half of the content showing at 25Hz. This could be quite easy to turn into 75Hz material. What do you think?

Also, movies are 24Hz, but that is still converted to 50 or 75Hz simply by speeding things up by 4%.

[NicolasB]

Originally Posted by Snaques

I may be wrong, but isn't video material interlaced

Yes.

Originally Posted by Snaques

and with half of the content showing at 25Hz.

No.

Originally Posted by Snaques

This could be quite easy to turn into 75Hz material. What do you think?

I think that you don't understand interlaced video.

A 50Hz source derived from film is, as you say, actually 25 progressive frames per second with each frame split in two: each pair of odd and even fields are both taken from the same frame, or, to put it another way, the odd and even fields were both captured at the same moment in time. But with a video source you are not talking about 25 progressive frames being photographed, you really are talking about 50 separate interlaced fields being photographed every second: there is a gap of 1/50th of a second between the time when the odd field is captured by the camera and the time when the even field is captured; then another 1/50th of a second before the next odd field.

This means you can't convert 50Hz video into 25 progressive frames per second; if you try, you will end up with horrendous "combing" artefacts of the sort that deinterlacers produce when they misiagnose video as film. Correctly deinterlaced 50Hz video can only be deinterlaced to 50 distinct progressive frames per second, which cannot easily be converted to 75.

Originally Posted by Snaques

Also, movies are 24Hz, but that is still converted to 50 or 75Hz simply by speeding things up by 4%.

The reason why film might be an issue with scanning backlights has nothing to do with 50Hz sources, it's a 60Hz issue. If a film encoded at 60Hz is deinterlaced correctly, you reconstruct the original 24 frames per second at (virtually) the correct speed. This means that, to display properly, it needs to be shown at a multiple of 24Hz. 24 and 48 would be too slow (would produce visible flicker) so it would have to be 72, 96 or 120. 72 and 96 are not exact multiples of 60, so the TVs may not be built to work at those speeds. 120Hz would be fine for both 60Hz video and 60Hz film, but I remain a teensy bit sceptical as to whether the actual LCD panel can really switch fast enough for a 120Hz frame rate.

[Snaques]

Originally Posted by NicolasB

I think that you don't understand interlaced video.

A 50Hz source derived from film is, as you say, actually 25 progressive frames per second with each frame split in two: each pair of odd and even fields are both taken from the same frame, or, to put it another way, the odd and even fields were both captured at the same moment in time.

Whopsie daisy, I really made a mess of my thoughts there. I got Hz and sec mixed up, so there you go.

This means you can't convert 50Hz video into 25 progressive frames per second; if you try, you will end up with horrendous "combing" artefacts of the sort that deinterlacers produce when they misiagnose video as film. Correctly deinterlaced 50Hz video can only be deinterlaced to 50 distinct progressive frames per second, which cannot easily be converted to 75.

Now back to the issue.
As you said "the odd and even fields were both captured at the same moment in time". Now why couldn't you reconstruct 25 progressive frames? One interlaced field includes half of the information of one such frame. Since two consecutive fields are captured at the same moment, wont they include the info of the same 25p frame?

Then conserning the 120Hz technique. I think there could be two solutions:
1. The one planned for some 100Hz LCD TV's where every other frame is dark. This is achieved by dimming the backlight which is a lot faster than the LCD panel. This dimming hides the part of the sequence where the crystals are in transition between two states. The LCD panel still works at 50Hz (or 60vs120Hz).

Other solution would be driving the TV with the 60Hz signal and the TV would estimate images between every signal image doubling the refresh rate to 120Hz. I think that these intermediate frames are usually on the transition path of the crystal and therefore the response time of, say 6ms, would be sufficient. It would also reduce the sensation of afterglow by adding fluidity to the scene. This kind of aproach was used in Samsung's first 100Hz LCD TV:
http://www.behardware.com/articles/6...afterglow.html

I would really like to know more about the Samsung's MPA technology. It's supposed to combine 120Hz refresh and screening to reduce afterglow.

[cerebros]

Originally Posted by Snaques

Now back to the issue.
As you said "the odd and even fields were both captured at the same moment in time". Now why couldn't you reconstruct 25 progressive frames? One interlaced field includes half of the information of one such frame. Since two consecutive fields are captured at the same moment, wont they include the info of the same 25p frame?

Because NicholasB was talking about 50Hz video, not 50Hz derived from film. With the 50Hz video each field is captured 1/50th of a second apart, therefore there will be movement between fields resulting in the combing artefacts mentioned if you try and convert it to 25p.