or Connect
AVS › AVS Forum › Display Devices › Display Calibration › Plasma Owners - How do you calibrate saturation?
New Posts  All Forums:Forum Nav:

Plasma Owners - How do you calibrate saturation? - Page 3

post #61 of 94
yes, I understand the measurement now. I can use the luminance sweeps on the GCD disk to do this. Here are the results at 100% saturation and measured at 25,50,75, and 100% luminance for each color.





ScreenShot2012-06-10at13830PM.png


Of course what you can see is that the onboard calibration system allows us to better distribute the inherent non-linearities compared to the example you plotted above. This happens as a result of minimizing perceptual errors in the calibration process. By the way, this has nothing to do with ABL, the colors are coupled due to the physical properties of the phosphor emission. In any case the onboard calibration systems allow for very accurate calibration over large portions of the gamut, at least on this display. The main point is that you can't just look at a plot like this and say that plasmas can't reproduce colors correctly without a 3D-LUT system, you have to measure the perceptual errors over your region of interest to make that kind of statement.


Color calibration results for the above display:


Average dE94=0.92, max=1.44
181


Tracking chart at 75% luminance - 75L/100S Average dE94=1.4, max=2.1
400
Edited by zoyd - 6/10/12 at 11:17am
post #62 of 94
Try the same with different size patches and you will see a different result... It's the ABL.

Plasma phosphor actually has good RGB Separation, as do old crt displays.

You can see this by the similarity of the RGB plots one against the other, rather than compared to the grey scale plot.

As a result any non 3D LUT calibration will only be valid for those exact patches. The moment different signals are used the calibration will be rather different.

The joys of plasmas!

wink.gif
post #63 of 94
If it helps (although its difficult to see in a single grab and not being able to rotate the cube) this is a plot of a 3d LUT for a plasma calibration - as you can see it is not a simple linear set of points, with complex distortions throughout the cube, as well as higher frequency changes.

The reduction in the 'area' of the cube towards the top also shows how saturation as to be reduced dynamically with image level - not something you can easily do without a 3D LUT.

352

Does this help?

A display with good RGB Separation (for what ever reason) is a lot easier to calibrate as the same display of the LUT would be far less 'complex'.

I'm also intrigued but the large variation in the individual RGB plots in your graph - we have never seen that on any plasma we have profiled for RGB Separation issues. That's a lot more like a D-ILA display...
post #64 of 94
Quote:
Originally Posted by Light Illusion View Post

Try the same with different size patches and you will see a different result... It's the ABL.
Plasma phosphor actually has good RGB Separation, as do old crt displays.
You can see this by the similarity of the RGB plots one against the other, rather than compared to the grey scale plot.
As a result any non 3D LUT calibration will only be valid for those exact patches. The moment different signals are used the calibration will be rather different.
The joys of plasmas!
wink.gif

You are not making this easy.rolleyes.gif Your first claim was that plasmas need your gizmo to be able to be calibrated accurately. I've demonstrated that is not the case for a common commercially available panel under one operating condition (1 set of stimuli). Now you say that doesn't count because my answer will change if I use different stimuli. Ok, if that is the case your solution will have the same problem, 2 different sets of stimuli will generate 2 different LUTs.

So, here is the same measurement using 10% windowed patterns. The first set above used 1% windowed patterns on a 25% fixed background. Note that these two cases offer dramatically different APL levels to the display. The first is fixed at 25% stimulus, the second varies from an 3% APL to 30% APL depending on color and level.

ScreenShot2012-06-10at50725PM.png

yes, there is a (small) difference, but the color errors for this differ from the case above by less than 1 dE94 across the board, i.e. imperceptible. I can go into great detail on why the difference between the two plots is not the ABL circuit (it's dynamic contrast at low APL). You'll notice that the functional form of the separation has not changed, all three curves have just shifted up a couple of percent, this is on EOTF effect.
Quote:
I'm also intrigued but the large variation in the individual RGB plots in your graph - we have never seen that on any plasma we have profiled for RGB Separation issues. That's a lot more like a D-ILA display...

With the internal 6x3 CMS system I can make any single luminance triplet match perfectly (zero separation). The process of calibrating to minimize perceptual error in the color set I have chosen causes gamut areas that do not contribute significantly to the dE calculations to be sacrificed in favor of those that do.
Edited by zoyd - 6/10/12 at 5:26pm
post #65 of 94
Quote:
Originally Posted by Light Illusion View Post

Try the same with different size patches and you will see a different result... It's the ABL.
Plasma phosphor actually has good RGB Separation, as do old crt displays.
You can see this by the similarity of the RGB plots one against the other, rather than compared to the grey scale plot.
As a result any non 3D LUT calibration will only be valid for those exact patches. The moment different signals are used the calibration will be rather different.
The joys of plasmas!
wink.gif
Can you provide a clear explanation of a test that is used to identify and measure "RGB separation"? What test pattern(s) are used, what are you looking for, and what constitutes passing or failing performance?
post #66 of 94
As I've interpreted it in the above plots the y-axis is (color luminance)/(rec.709 color ratio * peak white luminance) and the x-axis is white luminance/peak white.
Edited by zoyd - 6/10/12 at 5:30pm
post #67 of 94
Quote:
Originally Posted by zoyd View Post

As I've interpreted it in the above plots the y-axis is (color luminance)/(rec.709 color ratio * peak white luminance) and the x-axis is white luminance/peak white.
OK, straightforward enough. Then this just shows luminance errors of the primary colors at different levels of stimulus, right?
post #68 of 94
The correct way to plot RGB Separation is to pot the R, G, and B primary values relative to the equivalent grey scale value.

So, for a 50% value, for example, use a Gray patch that is a value of 128 (8 bits), and then pure Red at 128, pure Green at 128, and pure Blue 128.

The pure R, G and B measurements 'should' equal the same colour ratio as in the grey patch, based on the specific colour space profile.
This is a gamut plot, not luminance, which would explain your difference?

The plot should show the variation from the expected. I'm not sure that is what your plots are showing?
(It would actually bee good to profile the same exact display to see the results)

As I have said - a 3D LUT will get you the closest calibration - but that will not be perfect because of the problems we have been discussing.
But, 3D LUTs have the benefit of being able to alter any of the three colour outputs (which includes changing saturation separately to level) for any given input - so changing Red in can alter Green or Blue out, not just Red (actually not even Red if needed). This means you can plot for the various changes that are needed as input alters. This will get you 'closer' to accurate calibration.

There is info on this here: http://www.lightillusion.com/luts.html

Non 3D LUT based calibration will be less accurate overall, but with these problems neither will ever be fully accurate.

As I said before, we tell our professional clients to not use plasmas for any form of colour critical work.
Edited by Light Illusion - 6/11/12 at 3:05am
post #69 of 94
Quote:
Originally Posted by Light Illusion View Post

The correct way to plot RGB Separation is to pot the R, G, and B primary values relative to the equivalent grey scale value.

I'm assuming you mean measured luminance when you say primary and grey "values", is that correct?
Quote:
So, for a 50% value, for example, use a Gray patch that is a value of 128 (8 bits), and then pure Red at 128, pure Green at 128, and pure Blue 128.
The pure R, G and B measurements 'should' equal the same colour ratio as in the grey patch, based on the specific colour space profile.
This is a gamut plot, not luminance, which would explain your difference?
The plot should show the variation from the expected. I'm not sure that is what your plots are showing?

To take your 50% example, this was done by displaying 128 bit grey and fully saturated red, green, and blue patterns at 128 bit amplitudes in sequence as you describe. Each color should have a specific luminance relative to white:

R/W=21.26%
G/W=71.52%
B/W=7.22%
W/W=100%

I've plotted the ratio of the measured luminance to those expected values and then normalized to peak white so everything fits on the same scale.
Quote:
(It would actually bee good to profile the same exact display to see the results)
As I have said - a 3D LUT will get you the closest calibration - but that will not be perfect because of the problems we have been discussing.
But, 3D LUTs have the benefit of being able to alter any of the three colour outputs (which includes changing saturation separately to level) for any given input - so changing Red in can alter Green or Blue out, not just Red (actually not even Red if needed). This means you can plot for the various changes that are needed as input alters. This will get you 'closer' to accurate calibration.
There is info on this here: http://www.lightillusion.com/luts.html
Non 3D LUT based calibration will be less accurate overall, but with these problems neither will ever be fully accurate.
As I said before, we tell our professional clients to not use plasmas for any form of colour critical work.

I understand how the 3D LUTs work but for my system there would be no added value in using such a system if I am already measuring color accuracy errors below perceptual limits.

I've demonstrated the following with this example:
1. Plotting color separation tells us nothing about the level of perceptual error one can expect from the display. This can only be obtained through actual measurement.
2. One can obtain a color calibration on my display that will be perceptually indistinguishable from that obtained using a 3D-LUT over the area of the gamut enclosed by the first 12 Munsell color checker patches and the fully saturated 75% amplitude boundary. This has been verified through measurement using two pattern sequences, one with 10% by area windows and one with 1% by area windows on a 25% grey background.
post #70 of 94
Hey, if you like what you have, then I'm not going to argue!

All calibration is, at the end, about having something that works for you.

But, with your calibration, changing the size of the calibration patches will result in a different set of results.
While that is true with a 3D LUT calibration, the variation will be a lot less.
But, still not 'accurate' in the true sense of the word.

No display that has poor RGB Separation can ever be calibrated 'accurately' without the use of a free-form 3D LUT.
This is a simple fact.

But, that fact is based on 'accuracy' being 'measured'.
As above, if the result of ANY calibration technique is acceptable, that's often good enough.

With the plasmas I calibrated the other day (the data I have used here) we actually ended up not using the 3D LUTs as the 3D LUT box the customer had (BMD HDlink Pro) distorts the signal being passed through them, so making the end result less accurate than without... Using a different LUT box (a Davio we have in our company tools) the result was much better, but the cost prohibitive for the client's three plasmas.

At the end of the day, I hope the information I have provided helps understand the issues with plasma calibration.

The info can be used or ignored - makes no difference to me wink.gif

But, the information is accurate.
post #71 of 94
Quote:
Originally Posted by Light Illusion View Post

No display that has poor RGB Separation can ever be calibrated 'accurately' without the use of a free-form 3D LUT.
This is a simple fact.

Without quantifying "poor" and "accurate" it is literally impossible to evaluate that statement's validity. You can call it a belief if you'd like but not a fact.

What it boils down to is that we have two different definitions of accurate. Your definition appears to be zero RGB separation. I define accurate to mean below the limits of human perception as quantified by the CIE. Your definition is based on a bottom-up engineering principle(eliminate all sources of error) while mine is based on a quantifiable end-user requirement (are any remaining errors going to be noticeable).



edit: A quick follow-up regarding color grading metrics. In the example presented above I used CIE dE94 to gauge perceptibility and I've assumed that if the average is below 1 and max is below 1.5 there is no point in further optimization. There is another metric one can use defined in EBU tech memo 3325 which uses the difference in u'v' coordinates between target and reference. So I thought it would be interesting to calculate results using that metric. That memo specifies 18 color patches of which the 12 Munsell colors I've used are a subset. To meet the criteria of a grade 1 studio monitor in the color reproduction section, the error radius must be equal to or less than 0.003. Of the 12 patches I've used, all of them are within 2 radii, 10 are within 1.5 radii, and 7 of them are within 1 radius. The average error radius = 1.
Edited by zoyd - 6/11/12 at 10:56am
post #72 of 94
It is very true that our clients do demand a very high level of accuracy - its a major requirement for post-production operations.
LightSpace CMS defines 'poor' or not as being with acceptable limits for our client's expectations...

As such, the point about poor RGB Separation is fact (actually its a fact regardless, but you may choose to accept a different set of 'acceptable' parameters).

The problem is that without working to a set of accurate parameters there are no guarantees. For home use that is probably acceptable - for out customers it is not.

But, as I said, I provide this as information - use it or not. Doesn't bother me at all - but I hope it's of some interest biggrin.gif

The bottom line for us is that plasmas can never be calibrated to an acceptable level of accuracy due to ABL - for our client's use.
But where they have them we calibrate as best we can - and we need free-form 3D LUTs for that.

And don't get me started on D-ILA projectors - we have had a lot of clients forced to reject them as not being fit for purpose... mad.gif
post #73 of 94
Quote:
Originally Posted by Light Illusion View Post

It is very true that our clients do demand a very high level of accuracy - its a major requirement for post-production operations.
LightSpace CMS defines 'poor' or not as being with acceptable limits for our client's expectations...
As such, the point about poor RGB Separation is fact (actually its a fact regardless, but you may choose to accept a different set of 'acceptable' parameters).
The problem is that without working to a set of accurate parameters there are no guarantees. For home use that is probably acceptable - for out customers it is not.
But, as I said, I provide this as information - use it or not. Doesn't bother me at all - but I hope it's of some interest biggrin.gif
The bottom line for us is that plasmas can never be calibrated to an acceptable level of accuracy due to ABL - for our client's use.
But where they have them we calibrate as best we can - and we need free-form 3D LUTs for that.
And don't get me started on D-ILA projectors - we have had a lot of clients forced to reject them as not being fit for purpose... mad.gif

Hi Steve,

Very well said. I do a fair amount of calibration work for post production houses here in NYC. I am currently working with SpectraCal on the beta program for their Studio Version of CalMAN Pro software. This new version will have a 3D LUT builder in it for precisely the reasons you state above. It is an interesting conundrum of ABL in plasma panels versus the really poor black level of the broadcast LCD monitors from JVC and others.

As for front projection systems I am not a fan of LCoS (DILA) projection display technology. DLP is the way to go for really big screen applications.
post #74 of 94
Quote:
Originally Posted by TomHuffman View Post

OK, straightforward enough. Then this just shows luminance errors of the primary colors at different levels of stimulus, right?

sorry, missed the question - yes, that's right.
post #75 of 94
Quote:
Originally Posted by Monitorman View Post


Hi All,


Luminance of the colors is more important than the actual x and y coordinate of the color especially if it is too low..

You can't say that, it's a matter of degree. A 30% color luminance error is much less obvious than a situation where Green is displaced (in xy or uv coordinates) half-way to Yellow on the Rec709 triangle (which is what I'm seeing right now on a projector that will remain nameless).
post #76 of 94
Quote:
Originally Posted by ndaa75 View Post


Thank you Doug for a very comprehensive explanation.

Is [not being able to find correct adjustments for color luminance and saturation at the same time] a common issue for plasma units?

Also, I've read many about many different colour fields to use for calibration - some say use large APL patters or small APL patters (such as those on the AVSHD709 disk), others say the standard window format is fine.

What is the general recommendation here for colour calibration on plasma TV's?

I wouldn't say it's "common" per se, but it does happen and it may happen with one model in a product line but not another. I wouldn't say it's uncommon either. It depends on the manufacturer and how well they've chosen phosphors, how accurate the underlying driving technology is, how good or bad their CMS controls are, their operating firmware, and any number of other variables.
post #77 of 94
Quote:
Originally Posted by kyokushinkai View Post

Hi Doug.


I have a problem with this topic. I have a JVC X-30 and calibrate I encountered this problem.

I have never used an X30 or RS40 projector before. But I have a lot of time working with the RS50 & RS60 (X50 and X70). If the X30 controls are like the X50/70 controls, you can't get "perfect" calibration because the controls do not work right. I have spent many hours using the "graph" tool in the RS50/60 to try to make gamma and color accurate for the whole grayscale and cannot do it. The graph controls LOOK like they should work, but they do not. You would think if there were separate controls for white, red, green, and blue that you could fix any color problem at any part of the grayscale, but you cannot.

The CMS controls on the RS50/60 were also not capable of producing perfect calibration... they will improve some of the problems, but they cannot fix all of the problems. If the X30 has the same controls and same limitations, it's not surprising that you are having trouble getting everything to be good at the same time. I would say that I can get a good-but-not-perfect calibration with the RS50/60 controls. The controls for the RS55/65 are MUCH better and work as you would expect them to work. You can get excellent results for grayscale and CMS. For the RS50/60 the only way I found to get a "perfect" calibration was to use a Lumagen Radiance video processor.
post #78 of 94
Quote:
If the X30 controls are like the X50/70 controls, you can't get "perfect" calibration because the controls do not work right.

See my comments previously regarding D-ILA technology.
It has very poor RGB Separation, and as a result can unfortunately be near impossible to calibrate without the use of free-form 3D LUTs
(Something that can alter a given output based on any input - see here: http://www.lightillusion.com/luts.html)

As I comment above, we have had a number of our customer return their D-ILA systems as not being fit for purpose.

Plasmas can be bad enough to calibrate - D-ILA technology takes it to a whole new level mad.gif
post #79 of 94
Quote:
Originally Posted by Light Illusion View Post

Quote:
If the X30 controls are like the X50/70 controls, you can't get "perfect" calibration because the controls do not work right.
See my comments previously regarding D-ILA technology.
It has very poor RGB Separation, and as a result can unfortunately be near impossible to calibrate without the use of free-form 3D LUTs
(Something that can alter a given output based on any input - see here: http://www.lightillusion.com/luts.html)
As I comment above, we have had a number of our customer return their D-ILA systems as not being fit for purpose.
Plasmas can be bad enough to calibrate - D-ILA technology takes it to a whole new level mad.gif

how well do LCDs calibrate?
post #80 of 94
Quote:
Originally Posted by Light Illusion View Post

Quote:
If the X30 controls are like the X50/70 controls, you can't get "perfect" calibration because the controls do not work right.
See my comments previously regarding D-ILA technology.
It has very poor RGB Separation, and as a result can unfortunately be near impossible to calibrate without the use of free-form 3D LUTs
(Something that can alter a given output based on any input - see here: http://www.lightillusion.com/luts.html)
As I comment above, we have had a number of our customer return their D-ILA systems as not being fit for purpose.
Plasmas can be bad enough to calibrate - D-ILA technology takes it to a whole new level mad.gif

The current-year D-ILA RS55/65 projectors calibrate very close to perfection with their user menu controls if you understand how to use them.

You cannot blame an entire display technology for anything when the technology isn't the problem... the technology MAY have characteristics and limitations, but the design/implementation of the available controls and how well or poorly the engineers implement critical imaging performance is every bit as important as the imaging technology itself.

One thing I can say without reservation... I've never, EVER, seen an LCD, D-ILA, or SXRD projector that can do 3D without ghosting/crosstalk issues. The $25K Sony 4K projector comes closest to being ghost-free, but it's not perfect. But EVERY 3D DLP projector I've ever seen does perfect ghost/crosstalk-freen 3D. That includes very inexpensive (like $1K) DLP projectors and very expensive DLP projectors (like $55k). I've had manufacturers send me LCD or LCoS projectors they thought were ghost-free, but I find problems they never noticed right away and every time, if they get a movie title I have here and they have there, I can show them 3D failures. That is a "real" technology limitation. As far as I can tell, calibration issues with D-ILA, SXRD, and LCD projectors are always caused by controls that don't do what they are supposed to do, or there's too much control interaction between controls. That's not a technology limitation, that's a design/development-budget limitation.
Edited by Doug Blackburn - 6/13/12 at 11:39am
post #81 of 94
Well, at home, and by choice, I have a Sony Bravia LCD...

And professionally, we recommend LCD for direct display use for grading - something like the Penta HD2 Line series displays.
LCDs actually calibrate very well, but do suffer other (smaller) problems.

Viewing angle can be poorer, and the black can be 'less black' than plasma, but then plasmas tend to be be very noisy, and have far less 'granularity' (ability to see detail steps).

As it happens, the best grading display available at the moment, for professional use, is an LCD - from Dolby - with the Pentas very close in ability.

So yes, good LCDs do calibrate very well.
post #82 of 94
Quote:
Originally Posted by Light Illusion View Post

Well, at home, and by choice, I have a Sony Bravia LCD...
And professionally, we recommend LCD for direct display use for grading - something like the Penta HD2 Line series displays.
LCDs actually calibrate very well, but do suffer other (smaller) problems.
Viewing angle can be poorer, and the black can be 'less black' than plasma, but then plasmas tend to be be very noisy, and have far less 'granularity' (ability to see detail steps).
As it happens, the best grading display available at the moment, for professional use, is an LCD - from Dolby - with the Pentas very close in ability.
So yes, good LCDs do calibrate very well.

interesting considering most HT enthusiasts prefer plasmas.... I prefer LCD since it doesn't have ABL and can handle ambient light very well (matte screen).
post #83 of 94
Quote:
Originally Posted by PlasmaPZ80U View Post

interesting considering most HT enthusiasts prefer plasmas.... I prefer LCD since it doesn't have ABL and can handle ambient light very well (matte screen).

I need plasma in two places because of viewing angle requirements.


C'mon OLED....
post #84 of 94
Quote:
The current-year D-ILA RS55/65 projectors calibrate very close to perfection with their user menu controls if you understand how to use them.

I have to disagree - I have attempted to calibrate many D-ILA projectors, and not one has been capable of accurate calibration - when balanced the gamut is incorrect, or when gamut is ok, the balance is incorrect. But, as said before our calibration requirements are very tight.
Quote:
One thing I can say without reservation... I've never, EVER, seen an LCD, D-ILA, or SXRD projector that can do 3D without ghosting/crosstalk issues.

And that I have to disagree with - ghosting is all to do with the 3D technology, not the display technology. If the separate eye images are projected without overlap, there is no ghosting in the image... but, if the 3D technology used cannot separate the images correct, then ghosting will be visible.

There tends to be more ghosting with Real-D for example, as well as with cheaper active systems.

The only time the display can cause ghosting is when there is a delay with screen image switching - causing the previous image to remain visible when the nest is displayed.
The switching time of most 'quality' display technology is below this threshold.

Light Illusion does a lot of Stereo 3D work, helping set-up post-production operations and production workflows for projects.
Quote:
or there's too much control interaction between controls
That is an effect of the technology - cross-talk between the colour channels - a definition of poor RGB Separation, and is the exact issue I was flagging.
post #85 of 94
Quote:
Originally Posted by Light Illusion View Post

But, as said before our calibration requirements are very tight.

Can you tell us what your pass/fail color grading metric is?
post #86 of 94
Quote:
Originally Posted by Light Illusion View Post

Quote:
The current-year D-ILA RS55/65 projectors calibrate very close to perfection with their user menu controls if you understand how to use them.
I have to disagree - I have attempted to calibrate many D-ILA projectors, and not one has been capable of accurate calibration - when balanced the gamut is incorrect, or when gamut is ok, the balance is incorrect. But, as said before our calibration requirements are very tight.
Quote:
One thing I can say without reservation... I've never, EVER, seen an LCD, D-ILA, or SXRD projector that can do 3D without ghosting/crosstalk issues.
And that I have to disagree with - ghosting is all to do with the 3D technology, not the display technology. If the separate eye images are projected without overlap, there is no ghosting in the image... but, if the 3D technology used cannot separate the images correct, then ghosting will be visible.
There tends to be more ghosting with Real-D for example, as well as with cheaper active systems.
The only time the display can cause ghosting is when there is a delay with screen image switching - causing the previous image to remain visible when the nest is displayed.
The switching time of most 'quality' display technology is below this threshold.
Light Illusion does a lot of Stereo 3D work, helping set-up post-production operations and production workflows for projects.
Quote:
or there's too much control interaction between controls
That is an effect of the technology - cross-talk between the colour channels - a definition of poor RGB Separation, and is the exact issue I was flagging.

If you can't calibrate an RS55 or RS65 to near perfection with the internal controls (and to essentially complete perfection with a Lumagen Radiance processor) you don't know how to use the controls or modes... My results with the RS65 projector I reviwed are as follows:

Grayscale dEuv range: 0.6 or lower for 20% to 100% and 1.7 at 10% and 20%
Gamma Range: 10%-90%= 2.29-2.31 (I used 2.3 for my target Gamma)
Color dEs CIE 1994 & Luv
Red: 0.4 & 4.4
Green: 1.3 & 4.8
Blue: 0.75 & 8.4
Cyan: 0.9 & 3.3
Magenta: 0.2 and 2.1
Yellow: 0.4 & 1.3

And as I said, the Lumagen Radiance processor will further reduce the dEuv errors to well below 1 or 2. It is NOT the display technology that keeps the dEuv errors from being lower, it is how the projector's controls work.If the projector technology was incapable of lower dEuv errors, the Lumagen Radiance processor would have no effect, but Radiance can be used to make dEuv errors smaller. So the display technology itself is not the limiting factor. If you are using dE94 as your standard, the dEs re already low enough without the Radiance processor to be below the threshold of human vision. If you cannot duplicate those results with the current JVC models (RS55 & 65, not sure about the 45), you aren't using the right settings and may not be using the available controls to their best advantage. I've achieved those results with a dozen RS65 projectors... none of them has been different by more than a few percent here or there.

I'd be VERY interested in knowing exactly WHAT brands and models of LCD/LCoS projectors you have seen that do "perfect" 3D because I have Blu-ray 3D scenes that ALWAYS fail on EVERY LCD/LCoS projector and NEVER fail on any 3D DLP projector. I'm talking about Blu-ray 3D content as sent to the projector with an Oppo BDP-93 Blu-ray disc player (used because it is one of the top 2 or 3 Blu-ray players in regards to accuracy). No LCD or LCoS projector has produced non-ghosted 3D on all the torture test scenes I use (Hugo, Meet the Robinsons, The Polar Express, scenes from IMAX Dinosaurs Alive!, IMAX Grand Canyon Adventure -- and there are many others, but those are my "go to" titles). I use those discs for every projector review and DLP projectors always produce perfect 3D on every scene that the LCoS and LCD projectors fail on. Furthermore, the DLP projectors never fail anywhere on any disc that I've tried with them. This is a display technology issue and it has ZERO to do wih the content on the disc. If the fault was the content on the disc(s), the DLP projectors would fail in the same places (they don't).
There was some early discussion that light leakage through the 3D glasses was the issue and not the projectors, but the appearance of "universal" 3D glasses has completely elminated that from being a factor, because the "universal" 3D glasses will fail in the torture test scenes when used with LCoS and LCD projectors, but the same universal glasses used with DLP projectors produce perfect results in the same scenes that fail on the LCD/LCoS projectors. Furthermore, I have some "Alpha test" color leakage patterns for 3D glasses that have to be matched to the gamma of the projector... once you have the right set of targets to match the projector's gamma, you can evaluate how much leakage there is in % using these patterns (leakage is analyzed "per color" and it is very rare for glasses to leak equally in red, green, and blue). 3D glasses leakage ranges from 0.2% to about 12% (worst I've see so far). Yet low leakage number glasses used with LCoS and LCD projectors still produce ghosted images in the torture test scenes while high leakage glasses used with DLP projectors still produce "perfect" non-ghosted 3D images. This makes it indisputable that the display technology IS a factor in 3D image quality.
Sony's new VPL-VW1000ES 4K projector is the first SXRD projector to have the SXRD imagers driven digitally (all previous SXRD imagers had the digital signal converted to analog just prior to getting to the imagers - this is straight from Sony engineers). The 1000ES has the least amount of ghosting I've ever seen froim an LCoS or LCD projector, but it's still not perfect and it does still fail on some of the torture test scenes though it does "pass" on some of the scenes that give other LCoS or LCD projectors problems. But every DLP projector I've used here has no problems with the scenes the 1000ES produces ghosting on. The 1000ES projector produced about 20 instances of ghosting over the entire running time of Hugo (for example), each one lasting an average of 2 seconds (primarily because there was an edit or camera motion that changed scene content enough that the ghosting no longer existed. That means there was about 40 seconds of "bad" 3D during Hugo on that particular projector. If you convert the running time of the movie to seconds and divide, you get a 3D "score" of 99.47%: meaning 99.47% of the time, there was no 3D ghosting or other 3D failure to integrate the image (this includes time when only a portion of the image had a ghosting problem). Other LCD and LCoS projectors have had "scores" in the range of 95%-99%, the JVC RS65 was about 98.7% if I recall correctly. Every DLP projector has had 100% scores.
Note that the score for every 3D movie is not necessarily the same. Hugo tends to have lower overall scores when there are problems with any given projector. The IMAX titles and Meet the Robinsons tend to have higher scores, though DLP projectors still score 100%. Captain America & Part 1 of Harry Potter and the Deathly Hallows also tend to have lower scores when there are failures, primarily because the images are not as bright & vibrant as the movies that are more troublesome for LCoS and LCD projectors. Many of the 3D errors in Meet the Robinsons happen the first time the flying time machine arrives in the future. The motion in the scene can make it difficult to see the ghosting problems, but pausing the image confirms the problems (again, DLP projectors produce perfect 3D in the same spots even when the movie is paused).

So, once again, saying you can't make an LCoS projector accurate is simply wrong (as shown with the data from JVC RS65 projector review) and saying the 3D content is the (ghosting) problem rather than the display tech is also wrong as shown by the data from Hugo and the Sony 4K projector, the best 3D LCoS projector seen to date. And using the same universal 3D glasses with both types of projector and getting different ghosting results also confirms that the problem is with the display tech and not some other variable.
Edited by Doug Blackburn - 6/14/12 at 11:22am
post #87 of 94
Quote:
Originally Posted by Doug Blackburn View Post

If you can't calibrate an RS55 or RS65 to near perfection with the internal controls (and to essentially complete perfection with a Lumagen Radiance processor) you don't know how to use the controls or modes... My results with the RS65 projector I reviwed are as follows:
Grayscale dEuv range: 0.6 or lower for 20% to 100% and 1.7 at 10% and 20%
Gamma Range: 10%-90%= 2.29-2.31 (I used 2.3 for my target Gamma)
Color dEs CIE 1994 & Luv
Red: 0.4 & 4.4
Green: 1.3 & 4.8
Blue: 0.75 & 8.4
Cyan: 0.9 & 3.3
Magenta: 0.2 and 2.1
Yellow: 0.4 & 1.3

I went through this same argument with LI earlier. He considers measurements that meet THX, ISF, and EBU guidelines for accuracy this way:
Quote:
Originally Posted by Light Illusion View Post

For home use that is probably acceptable - for out customers it is not.

And that's only a probably, mind you.
Edited by zoyd - 6/14/12 at 11:26am
post #88 of 94
Quote:
Originally Posted by Light Illusion View Post

Well, at home, and by choice, I have a Sony Bravia LCD...
And professionally, we recommend LCD for direct display use for grading - something like the Penta HD2 Line series displays.
LCDs actually calibrate very well, but do suffer other (smaller) problems.
Viewing angle can be poorer, and the black can be 'less black' than plasma, but then plasmas tend to be be very noisy, and have far less 'granularity' (ability to see detail steps).
As it happens, the best grading display available at the moment, for professional use, is an LCD - from Dolby - with the Pentas very close in ability.
So yes, good LCDs do calibrate very well.
The Penta HD2 LCDs have a contrast ratio of 1000-1500:1. With the nominal peak output of 85cd/m2, that's a MLL of 0.085-0.057 cd/m2. The current generation of consumer plasmas has a MLL in the 0.01-0.02 range. I am not a contrast freak, but this Penta black level not acceptable performance by any standard. For example, SMPTE's specification for contrast of a reference projector is 2000:1. The Dolby 42" LCD doesn't publish specs on contrast, but its MSRP is $39,995.

The requirements for professional broadcast displays and consumer displays are very different. The consumer market will tolerate relatively small errors in color accuracy in return for vastly better contrast--and in the case of plasmas--wider viewing angles.

I am not really disagreeing with anything you say here. But I am trying to provide some context.
Edited by TomHuffman - 1/6/13 at 12:48am
post #89 of 94
Quote:
Originally Posted by Doug Blackburn View Post

Color dEs CIE 1994 & Luv
Red: 0.4 & 4.4
Green: 1.3 & 4.8
Blue: 0.75 & 8.4
Cyan: 0.9 & 3.3
Magenta: 0.2 and 2.1
Yellow: 0.4 & 1.3
Doug, is the value for blue a typo? I can't imagine any color whose CIE94 is 0.75 with a CIELUV of 8.4. What is the xyY value?

Also, I think that the poster is referring to color accuracy throughout the entire color space, and not just at the gamut boundary.
post #90 of 94
As a final comment on the saturation/ABL issue (I think all other points have been done to death now smile.gif) the correct measurements for RGB Separation are as follows:

In a display with good RGB Separation, the X, Y, Z components of each primary colour measurement patch should be equal to the corresponding grey/white patch:

Xw = Xr + Xg + Xb
Yw = Yr + Yg + Yb
Zw = Zr + Zg + Zb

If this is not true, then the display has poor RGB Separation.

The plasma RGB Separation graph I posted shows that the intensity of the primary colour patch (R, G or B) is less than the intensity of the colour 'bit' of the corresponding grey/white patch.

The closer to the black line, the better RGB Separation the display has, and the easier to calibrate.

Hope this helps as a fuller explanation.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Display Calibration
AVS › AVS Forum › Display Devices › Display Calibration › Plasma Owners - How do you calibrate saturation?