or Connect
AVS › AVS Forum › Display Devices › Display Calibration › Gamma with normal windows or with APL?
New Posts  All Forums:Forum Nav:

Gamma with normal windows or with APL? - Page 2

post #31 of 193
Quote:
Originally Posted by LarryInRI View Post

The ones I use are not "nearly identical to one another." They are quite different. Read Ted's post above (#9). He uses "6 different Gammas for Rec.709: 2.18, 2.20, 2.22, 2.25, 2.27. 2.29".

There are definite visible changes with small gamma changes.


Larry

you are talking about power (formula) gammas, right?

if you do the math, a .02 point gamma change represents a minimal % change in luminance (even at the low end)... you might see a tiny difference when toggling between the two gammas but odds are you wouldn't know which was which if you weren't aware of which gamma was selected

based on what the pros have mentioned in the past here, targeting a difference of less than .05 doesn't make a significant difference.... so presets like 2.20, 2.25, 2.30, 2.35, 2.40 or 2.20, 2.30, 2.40 make sense but 2.18, 2.20, 2.22 is really just splitting hairs

that being said even the difference between 2.20 and 2.25 is small
Edited by PlasmaPZ80U - 8/15/13 at 8:32am
post #32 of 193
Quote:
Originally Posted by PlasmaPZ80U View Post

Quote:
Originally Posted by LarryInRI View Post

The ones I use are not "nearly identical to one another." They are quite different. Read Ted's post above (#9). He uses "6 different Gammas for Rec.709: 2.18, 2.20, 2.22, 2.25, 2.27. 2.29".

There are definite visible changes with small gamma changes.


Larry

you are talking about power (formula) gammas, right?

if you do the math, a .02 point gamma change represents a minimal % change in luminance (even at the low end)... you might see a tiny difference when toggling between the two gammas but odds are you wouldn't know which was which if you weren't aware of which gamma was selected

based on what the pros have mentioned in the past here, targeting a difference of less than .05 doesn't make a significant difference.... so presets like 2.20, 2.25, 2.30, 2.35, 2.40 or 2.20, 2.30, 2.40 make sense but 2.18, 2.20, 2.22 is really just splitting hairs

The difference is small in Luminance but Gamma Target Interacts to the Whole Color Saturations Inside the Cube also. It's a slight difference, you need to pause the screen to notice that 0.02-3 difference range.

This picture shows the difference of Target Luminance between Gamma 2.22 - 2.25 - 2.28:



Generally i'm using the memories to fix black level issues of Movie Mastering, To slight improve shadow detail or to reduce it.

The most imprortant is to generate this 6 memories it takes max 20 sec, to need to re-measure again my panel, using LightSpace.
Edited by ConnecTEDDD - 8/15/13 at 9:05am
post #33 of 193
Quote:
Originally Posted by LarryInRI View Post


Technically, why the difference? Which is correct -- the 4% normal windows or a corrected 10% with 22% APL?


Larry

Hi Larry, both are correct in the sense that you are calibrating the display to generate a D65 response to the two patterns. However, a 4% window with 30% stimulus on a black background yields an average driving luminance of 0.3^2.2*0.04=0.28% on a 2.2 gamma display. Your display will almost never operate at this APL with any source material (expect small windows rolleyes.gif ). Plasmas (or any display for that matter) are designed to operate with real video, so if you measure a difference between small windows and APL windows which simulate more realistic power loads, the APL type is the better technical choice.
post #34 of 193
Zoyd,

I thought I'd monitor power loads on my 65VT50 with one of those watt killer power strips that has an LCD on it that tells the amount of power used in real time along with a variety of other information.

I ran full screen patterns at 10% level up to 100% in 10% increments and also 25%.

What I found interesting was that 80% full screen drew more power than 90%. Still scratching my head on this one.

The other point was that a 25% screen still drew a lot less power than even a fairly dark scene when it came to real content.

What would you think would cause these differences?
post #35 of 193
is it true that recent plasmas have more aggressive ABL implementations than older ones (like the pioneers for example)? to meet stricter power consumption requirements?
post #36 of 193
Quote:
Originally Posted by ConnecTEDDD View Post

The difference is small in Luminance but Gamma Target Interacts to the Whole Color Saturations Inside the Cube also. It's a slight difference, you need to pause the screen to notice that 0.02-3 difference range.

This picture shows the difference of Target Luminance between Gamma 2.22 - 2.25 - 2.28:



Generally i'm using the memories to fix black level issues of Movie Mastering, To slight improve shadow detail or to reduce it.

The most imprortant is to generate this 6 memories it takes max 20 sec, to need to re-measure again my panel, using LightSpace.

that all makes sense then, especially since in your case having these six memories is not a big undertaking

it seems you agree that a .02-.03 point gamma difference is quite subtle and aside from watching a static screen, you'd need an A/B comparison to be able to fully appreciate the difference (also you'd want to watch a very dark scene with shadow detail just above black)
post #37 of 193
Quote:
Originally Posted by PlasmaPZ80U View Post


it seems you agree that a .02-.03 point gamma difference is quite subtle and aside from watching a static screen, you'd need an A/B comparison to be able to fully appreciate the difference (also you'd want to watch a very dark scene with shadow detail just above black)

There movies that need a 'slight' adjust of gamma to meet the falimiar picture you want from your setup, to fix some movie problem of slight black detail washout or some shadow detail lost so this slight difference in gamma helps in that specific situation.
post #38 of 193
Quote:
Originally Posted by PlasmaPZ80U View Post

is it true that recent plasmas have more aggressive ABL implementations than older ones (like the pioneers for example)? to meet stricter power consumption requirements?

ABL exists to reduce the size (people like thin flat panels) and cost of the display, not for energy savings although that is a by-product.
post #39 of 193
Quote:
Originally Posted by JimP View Post

Zoyd,

I thought I'd monitor power loads on my 65VT50 with one of those watt killer power strips that has an LCD on it that tells the amount of power used in real time along with a variety of other information.

I ran full screen patterns at 10% level up to 100% in 10% increments and also 25%.

What I found interesting was that 80% full screen drew more power than 90%. Still scratching my head on this one.

The other point was that a 25% screen still drew a lot less power than even a fairly dark scene when it came to real content.

What would you think would cause these differences?

I don't know how well wall power correlates with panel load but it is not surprising to me that a 25% screen is efficient since that is where the ABL transition typically occurs.
post #40 of 193
Thread Starter 
Quote:
Originally Posted by zoyd View Post

Hi Larry, both are correct in the sense that you are calibrating the display to generate a D65 response to the two patterns. However, a 4% window with 30% stimulus on a black background yields an average driving luminance of 0.3^2.2*0.04=0.28% on a 2.2 gamma display. Your display will almost never operate at this APL with any source material (expect small windows rolleyes.gif ). Plasmas (or any display for that matter) are designed to operate with real video, so if you measure a difference between small windows and APL windows which simulate more realistic power loads, the APL type is the better technical choice.


Hi, zoyd.

When using a black background, I don't understand the conclusion that you draw by modifying the basic input/output function with a linear factor of percent of window area. Indeed, as you show, using a 30% stimulus, a gamma of 2.2, and then multiplying it by the percent of the average driving luminance of the pattern area yields a small number. However, using your logic, for a 30% input stimulus, a 2.2 gamma, and this time with a 10% window we get an average driving luminance of 0.707%. This is still a small average driving luminance Again, using the original 30% stimulus, a 2.2 gamma, and increasing the window to full screen, ignoring ABL of course, using your pattern area multiplying factor we get the normally expected output of 7.07% of max output.

Looking at it from a different view, if the percent area is used as a constant multiplier to the input/output function, the results we get reflect the pattern size and simply reduce the expected luminance output by the selected percent area. Indeed, the average driving luminance is affected using a window. But that reduction will occur through the full range of the testing stimulus/luminance. That is confusing.

The input/output gamma transfer function is well known and describes the output luminance for a specific input stimulus. I do not grasp what you are conveying by reducing the output by the percent of window size when using a black background. Using the percent of area factor in the transfer function would lead me to want an window area large enough so that the max luminance of that larger window is just below the level where ABL kicks in.

What am I missing?


Sorry to continue this thread. I had planned to let it die since the subject has been hashed over many times in this forum.

Larry

typos edited
Edited by LarryInRI - 8/16/13 at 4:27pm
post #41 of 193
Quote:
Originally Posted by LarryInRI View Post

Indeed, as you show, using a 30% stimulus, a gamma of 2.2, and then multiplying it by the percent of the average driving luminance of the pattern area yields a small number.

I figure maybe Zoyd is suggesting that generally windows provide images with a considerably lower "average relative luminance" than would typically be found in a considerable portion of video content, such as movies, TV shows, etc. Whether or not using measurement images that are darker than typical content matters for calibration likely depends on how exactly an individual display operates. Clearly "average relative luminance" for calibration images doesn't matter at all for certain LCD models or settings, but I think some of the fluctuating brightness (FBr) discussions suggested that overly dark images might introduce possible measurement variations for certain plasmas.

Quote:
The input/output gamma transfer function is well known and describes the output luminance for a specific input stimulus.

For grayscale a transfer function basically describes the ideal gamma curve for the video information (picture levels), which is essentially just how various video levels should relate to each other, but the transfer function doesn't actually describe how luminance will necessarily measure for various video levels on a plasma display. The transfer function allows the sort of ideal "average relative luminance" calculation that Zoyd was doing, yet for some display types the actual measured on-screen luminance may tend to vary depending on other factors besides simply the ideal transfer function.

Quote:
Using the percent of area factor in the transfer function would lead me to want an window area large enough so that the max luminance of that larger window is just below the level where ABL kicks in.

I don't have a plasma to try to document how one individual model happens to operate, but generally the posted measurements I remember looking at seemed to suggest that certain plasmas may tend to measure luminance differently depending on factor(s) such as possibly the "average relative luminance" or the Average Picture Level (APL) of the displayed image. For example if you were to measure an increasing series of white windows, from the smallest measureable white window to a full field of white, the trend may be that the entire series of white measurements might generally result in decreasing luminance measurements as "average relative luminance" and Average Picture Level (APL) increases. For certain displays maybe there is a clear "level where ABL kicks in" as a white window increases in area, which could be tested with any software that provides adjustable window sizes, such as ColorHCFR, but personally I was never able to break down the information from plasma users in a way where this sort of idea clearly fit with the data.


"average relative luminance" - basically my intent was for this to mean running the video information through the transfer function and then averaging the image
Edited by alluringreality - 8/18/13 at 1:12am
post #42 of 193
Thread Starter 
Thanks, alluring. But I do understand, in an engineering sense, all the terminology and the simple transfer function. I am curious about the methodology, interpretation, and conclusion derived from the numerical value that zoyd characterized in his example. I understand how he is modifying the luminance output by the percent area of the window when determining a specific point in the gamma point curve. However, I do not understand his conclusion and I hope I stated my reasons for that confusion clearly -- but maybe I didn't.

Knowing zoyd, he will explain his rational. Knowing zoyd, he probably has a valid engineering point -- and maybe it will take a bit more explaining for me to grasp his reasoning.

Larry
post #43 of 193
Quote:
Originally Posted by JimP View Post


The gamma derived from using a large window on the plasma was a lot closer to the LCD's "look" than the 5% APLs. Deep shadows were a touch darker on the plasma than on the LCD. That would be simple enough to correct.. In comparison, the APL pattern calibrations looked washed out and lacked a certain fidelity to them which makes sense and shows up moreso when doing a side by side comparison. I never really noticed it with the plasma by itself.

Now the real question is do I really like it that way.

Well.....it's growing on me.

You are way oversimplifying what is going on. You CANNOT put a plasma and LCD side by side and watch 5 minutes of content and learn much of anything because the plasma has brightness limiting and the LCD does not and you CANNOT stop the plasma from doing that. No amount of calibration can stoo it and the plasma is NOT going to look more or less similar to the LCD when you consider images of ALL APLs... that liast bit is CRITICAL to understand, remember, commit to memory and to never underestimate. Because, all you do by using constant APL patterns is calibrate the plasma to a SINGLE APL but the plasma does not OPERATE on a single APL, the plasma's APL changes constantly. So the "look" of the images changes constantly as APL changes and you can't stop that and you can't say that either calibration method makes a plasma look more or less the way an LCD looks displaying the same content. I understand wanting to be able to reduce the problem to some identifiable quantity by comparing plasma to LCD (or any number of other comparisons or tests you might think up), but you can't. And trying to do so is just an exercise in futility as is this entre thread and all the threads discussing the best way to calibrate plasma (i.e. constant pattern size with black background or constant APL patterns, etc.).

Somewhere somebody calculated the long-term APL that a variety of TV and movie content delivers over months or even years of TV viewing. I can't find the reference but I seem to recall it being somewhere in the 40-50% range. If you could find that long term average and use that as your constant APL reference, you'd get as close as you could get with the constant APL method for calibrating plasmas. But each % you deviate from that long term average is going to look different on the plasma. And by the way, different brands and models of plasma have different ABL behavior so you wouldn't necessarily want to use the same APL patterns on every plasma if you were truly trying to do a constant APL calibration. You'd have to characterize the ABL performance of the plasma before you begin the calibration and select a specific APL value for the patterns that best keeps that plasma in the middle of it's performance range.

Trying to convince yourself that one method somehow looks better or worse than the other method is just an exercise in futility. You can find flaws with both methods and those flaws are not larger or smaller in magnitude with one method versus the other.
post #44 of 193
This picture is coming from Cine-Tal Davio Plasma Calibration Workflow (Proffesional World)





This Window Pattern Size of 25-40% they are suggesting for Panasonic Professional Series Plasma's, I got the same size suggestion from THX CineSpace Support for Plasma Profilings, when i was using THX CineSpace Evaluation Software.
post #45 of 193
Thread Starter 
Quote:
Originally Posted by HDTVChallenged View Post

Actually, something looks quite wrong with that graph compared to what I get on the final (unreleased) version of my spreadsheet. BT1886 should "bottom out" at ~2.19 at 5% stimulus ... decreasing even more (brighter) below .... Edit to come.

Yeah, you're right. I plugged in the wrong MLL for the previous chart. Here is the corrected one for 100 cd/m^2 and 0.005 cd/m^2. I still don't like the picture that the bt.1886 gamma curve yields.



Larry
post #46 of 193
Quote:
Originally Posted by Doug Blackburn View Post

Somewhere somebody calculated the long-term APL that a variety of TV and movie content delivers over months or even years of TV viewing. I can't find the reference but I seem to recall it being somewhere in the 40-50% range.

While I haven't seen measurements that clearly tie plasma variation to either "average relative luminance" or APL, the Digital Video Essentials website lists 15% and 35% APL over time.

http://www.videoessentials.com/glossary.php
post #47 of 193
Quote:
Originally Posted by LarryInRI View Post

I am curious about the methodology, interpretation, and conclusion derived from the numerical value that zoyd characterized in his example.

The underlying assumption in the comment from Zoyd seems to be that the plasma likely changes RGB balance roughly in line with average luminance. I'm not aware of people specifically documenting if measured changes in RGB balance on plasma relate well with average luminance, but the average luminance of typical video content over a long period of time is generally going to be closer to the average luminance of a 22% APL pattern than a 4% window run. For a majority of the measurement run the overall light output from the display will be brighter using a 22% apl pattern than a 4% window pattern, so the 22% APL pattern is generally closer to typical video content brightness. As a third party observer that doesn't own a VT60, I can say that calibrating RGB balance for the second chart would seem to cause the first chart to increase in blue across the board, which supports the "visible blue tint" comment from the first post. If the display does generally vary RGB balance roughly according to average luminance, I suppose the blue tint might apply more to darker than average scenes. If the blue tint also applied to scenes with a similar average luminance to the 22% APL measurement pattern, that would go against the idea of the RGB balance varying roughly in line with changes in average luminance. Realistically if you want to get a better idea of what exactly is going on with the display's RGB balance it makes sense to measure a lot of patterns, and while I don't always agree with zoyd on conclusions, the amount of measured data zoyd posts is usually rather impressive.
Edited by alluringreality - 8/18/13 at 7:59pm
post #48 of 193
Thread Starter 
I just spent the entire day re-reading the following threads:
http://www.avsforum.com/t/1441407/grayscale-variances-different-patterns-plasma-calibration
http://www.avsforum.com/t/1414313/better-to-have-2-2-gamma-or-stable-2-3-gamma-on-a-plasma
http://www.avsforum.com/t/1446386/abl-effects-measured-for-comparison
http://www.avsforum.com/t/1406352/gcd-gamut-calibration-disk

Things are somewhat more clear reading them in one big chunk rather than post-by-post on a daily basis. So I feel that now I have a better handle on zoyd's point. I doubt though that I will ever have the time or the patience to reproduce his voluminous data filled experiments. smile.gif

I am, however, going to recalibrate my VT60. This time I am will be setting the Pro1 mode using a 6.5 normal window and setting the Pro2 mode using a 10% window on a 22% APL screen. Gamma will be adjusted to be a flat 2.25 for both modes. Max luminance will be adjusted to be visibly the same by an iterative process of re-calibrations on one of the modes. That in itself will require enough of my energy. Which ever looks best to me after viewing each for a week or so is the one that I will settle on.

Any suggestions before I start will be appreciated.


Larry
post #49 of 193
Quote:
Originally Posted by LarryInRI View Post

Yeah, you're right. I plugged in the wrong MLL for the previous chart. Here is the corrected one for 100 cd/m^2 and 0.005 cd/m^2. I still don't like the picture that the bt.1886 gamma curve yields.

Larry

Yeah ... that looks more like it should. smile.gif I do find it "interesting" and a bit puzzling that you don't like it. FWIW, it's somewhat different result on my LCD. (Try BL = 0.122 Nits, WL = 117 Nits. ... which tends to hug the P.L. gamma =2.2 curve)

I would think that one would want to take advantage of the lower MLL and the (theoretically) higher contrast ratio on the VT60? Then again, perhaps the ABL issue negates all of that?

All I can say is that, last night, I reconstructed the BT1886 calibration on my LCD. I've been trying to pick my lower jaw up off of the floor ever since. smile.gif The thing has no business looking as good as it does right now.
Edited by HDTVChallenged - 8/19/13 at 10:43am
post #50 of 193
Quote:
Originally Posted by Doug Blackburn View Post

You are way oversimplifying what is going on. You CANNOT put a plasma and LCD side by side and watch 5 minutes of content and learn much of anything because the plasma has brightness limiting and the LCD does not and you CANNOT stop the plasma from doing that. No amount of calibration can stoo it and the plasma is NOT going to look more or less similar to the LCD when you consider images of ALL APLs... that liast bit is CRITICAL to understand, remember, commit to memory and to never underestimate. Because, all you do by using constant APL patterns is calibrate the plasma to a SINGLE APL but the plasma does not OPERATE on a single APL, the plasma's APL changes constantly. So the "look" of the images changes constantly as APL changes and you can't stop that and you can't say that either calibration method makes a plasma look more or less the way an LCD looks displaying the same content. I understand wanting to be able to reduce the problem to some identifiable quantity by comparing plasma to LCD (or any number of other comparisons or tests you might think up), but you can't. And trying to do so is just an exercise in futility as is this entre thread and all the threads discussing the best way to calibrate plasma (i.e. constant pattern size with black background or constant APL patterns, etc.).

Somewhere somebody calculated the long-term APL that a variety of TV and movie content delivers over months or even years of TV viewing. I can't find the reference but I seem to recall it being somewhere in the 40-50% range. If you could find that long term average and use that as your constant APL reference, you'd get as close as you could get with the constant APL method for calibrating plasmas. But each % you deviate from that long term average is going to look different on the plasma. And by the way, different brands and models of plasma have different ABL behavior so you wouldn't necessarily want to use the same APL patterns on every plasma if you were truly trying to do a constant APL calibration. You'd have to characterize the ABL performance of the plasma before you begin the calibration and select a specific APL value for the patterns that best keeps that plasma in the middle of it's performance range.

Trying to convince yourself that one method somehow looks better or worse than the other method is just an exercise in futility. You can find flaws with both methods and those flaws are not larger or smaller in magnitude with one method versus the other.

You really worked hard to miss the point of my post.

.....and where did you get the idea that I only spent 5 minutes making the comparison? Since you totally fabricated that information, I question if you are deliberately trying to misdirect the enthusiast community. I spent several days making these comparison where I paused a couple of hundred scenes to draw my conclusions. I was originally pointed in this direction by a pro calibrator that uses large windows on a VT50 due to how the algorithms work. I wanted to confirm this as the whole "use APL patterns" makes more sense to me. We often get conflicting information on fairly significant issues in these calibration threads and I wanted to see for myself which is the most correct. With this method, I can at least make the conclusion that when I calibrate to a gamma target of 2.2, then that's what I'm actually seeing on the display. Calibrate to a gamma target of 2.2 using APL windows, you're not getting 2.2. Whether or not a 2.2 gamma provides you the look that you want is a separate question..

I'll try to make this as brief as possible to not leave room for misinterpretation.

When comparing a display which has no abl issue to a plasma that does such as the VT50, using a large window pattern on the plasma will provide you with a closer match than using APL patterns. High APL scenes can't be compared due to the brightness limiting of plasmas. You can however make comparisons for mid to low APL scenes. In terms of gamma, large window pattern will be a much closer match on a VT50 than 5% APL patterns with 10, 18, 25% surrounds . This gives you a reference point for 2.2 gamma and what it's suppose to look like within the context of the limitations of a plasma, mainly the dimming of bright scenes..

I would agree that this is counter intuitive but I can't really argue with the results.

Like many of you, I've struggled with which pattern should I use. One pro calibrator will insist on APL patterns. Another will equally insist on large window patterns and their arguments are equally compelling. At a minimum, if you're serious about calibration which I think you would be otherwise you wouldn't be on these threads, try the comparisons for yourself. Theory only goes so far, then you have to do your own testing.

(Flamesuit on) smile.gif
post #51 of 193
Quote:
Originally Posted by HDTVChallenged View Post

Yeah ... that looks more like it should. smile.gif I do find it "interesting" and a bit puzzling that you don't like it. FWIW, it's somewhat different result on my LCD. (Try BL = 0.122 Nits, WL = 117 Nits. ... which tends to hug the P.L. gamma =2.2 curve)

I would think that one would want to take advantage of the lower MLL and the (theoretically) higher contrast ratio on the VT60? Then again, perhaps the ABL issue negates all of that?

Quoting myself to clarify that last statement a bit. On the VT60, BT1886 *will* result in a "darker" gamma response below 20%(ish) stimulus than P.L. @ 2.2, and if you add the effects of ABL on top of that, I can "see" (in theory) why that might make things look a wee bit murkier especially if you've been using PL @ 2.2. It may be that PL is a better choice for plasma ... I haven't had the opportunity to experiment with one here in the bunker.

But, that would be a bit ironic for the tech that is supposedly the "most like" good old CRTs, no? wink.gif
post #52 of 193
Quote:
Originally Posted by LarryInRI View Post

Thanks, alluring. But I do understand, in an engineering sense, all the terminology and the simple transfer function. I am curious about the methodology, interpretation, and conclusion derived from the numerical value that zoyd characterized in his example. I understand how he is modifying the luminance output by the percent area of the window when determining a specific point in the gamma point curve. However, I do not understand his conclusion and I hope I stated my reasons for that confusion clearly -- but maybe I didn't.

Knowing zoyd, he will explain his rational. Knowing zoyd, he probably has a valid engineering point -- and maybe it will take a bit more explaining for me to grasp his reasoning.

Larry

Hi Larry,

Sorry for the delayed response but I've been traveling a lot lately. alluringreality captured very well the point I was making and I'm happy to see that the Poynton quote follows very closely the luminance levels I have found to be good targets for the "test as you fly" calibration philosophy. It also seems you are now OK with how the math is done? The plasma ABL circuit reacts to the total energy required to display an image. The total energy is proportional to the pixel duty cycle determined by input stimulus x total number of pixels at that stimulus. Using the area factor yields the relative average luminance over all pixels.

And my point was not specifically that RGB tracking or anything else follows the ABL response the same way for plasmas in general, but that if you actually measure a difference between windowed patterns and APL type patterns, that APL patterns have a theoretical advantage over windowed and are more "correct" in that sense. Windowed patterns have little in any theoretical basis (that I have seen) but they are easy to make and give good results.
Edited by zoyd - 8/19/13 at 12:12pm
post #53 of 193
Thread Starter 
Thanks, zoyd. Yes, it is all pretty clear now. I did get hung up for a while on a few of your charts -- as did Tom and Alluring so I didn't feel too badly about it. smile.gif

Of particular interest was the data and discussion concerning the chromaticity shift in the 20% to 40% stimulus range when comparing standard windows on a black background and windows on a constant APL background. I think this finding deserves more consideration.

Larry
post #54 of 193
Quote:
Originally Posted by JimP View Post

You really worked hard to miss the point of my post.

.....and where did you get the idea that I only spent 5 minutes making the comparison? Since you totally fabricated that information, I question if you are deliberately trying to misdirect the enthusiast community. I spent several days making these comparison where I paused a couple of hundred scenes to draw my conclusions. I was originally pointed in this direction by a pro calibrator that uses large windows on a VT50 due to how the algorithms work. I wanted to confirm this as the whole "use APL patterns" makes more sense to me. We often get conflicting information on fairly significant issues in these calibration threads and I wanted to see for myself which is the most correct. With this method, I can at least make the conclusion that when I calibrate to a gamma target of 2.2, then that's what I'm actually seeing on the display. Calibrate to a gamma target of 2.2 using APL windows, you're not getting 2.2. Whether or not a 2.2 gamma provides you the look that you want is a separate question..

I'll try to make this as brief as possible to not leave room for misinterpretation.

When comparing a display which has no abl issue to a plasma that does such as the VT50, using a large window pattern on the plasma will provide you with a closer match than using APL patterns. High APL scenes can't be compared due to the brightness limiting of plasmas. You can however make comparisons for mid to low APL scenes. In terms of gamma, large window pattern will be a much closer match on a VT50 than 5% APL patterns with 10, 18, 25% surrounds . This gives you a reference point for 2.2 gamma and what it's suppose to look like within the context of the limitations of a plasma, mainly the dimming of bright scenes..

I would agree that this is counter intuitive but I can't really argue with the results.

Like many of you, I've struggled with which pattern should I use. One pro calibrator will insist on APL patterns. Another will equally insist on large window patterns and their arguments are equally compelling. At a minimum, if you're serious about calibration which I think you would be otherwise you wouldn't be on these threads, try the comparisons for yourself. Theory only goes so far, then you have to do your own testing.

(Flamesuit on) smile.gif

Still baloney. Constant APL patterns exercise ONESINGLE point in the plasma ABL spectrum. You cannot stop ABL when viewing images on plasmas, but constant APL windows also produce a calibration at a single/constant ABL which will never happen when you view content on a plasma. Your calibration should include at least SOME range of ABL action on the part of the plasma TV and the only way to do that is with traditional window patterns with a black background.

Avoiding critical evaluations of high APL images against the LCD "reference"... OK, in a practical sense you can't do that comparison. But that implies being rather cavalier about the appearance of images when the plasma's ABL is getting more significant... if you only care how the low end of the luminance range looks, (as that's what you are comparing to the LCD), how will you ever know if your constant APL calibration patterns are doing justice to the plasma calibration when ABL begins to be more of a factor? You won't. Which is why you really need to have SOME ABL action in your calibration. I'm no advocating intentionally pumping up the ABL, but I am advocating not eliminating it entirely from the calibration (as you do when using constant APL patterns).

I have NO IDEA what you mean by "large patterns" If you mean "full screen" or anything larger than 10%-12% of the physical size of the screen, that's just a red herring, nobody (I know) would advocate calibrating a plasma display with full screen or even 20% or larger window patterns. If you are talking about 5% physical windows as being "proper" and 10% windows as being "large".... I suggest using something more descriptive. Like "10%=size".

And by making 1 comparison, you can't make any sweeping conclusions because, as mentioned in the previous post, different brands and models of plasma have different ABL "action" so what you perceive as working well for 1 particular plasma, could be a bad choice for some other plasma.
post #55 of 193
Quote:
Originally Posted by JimP View Post

Snip
Like many of you, I've struggled with which pattern should I use. One pro calibrator will insist on APL patterns. Another will equally insist on large window patterns and their arguments are equally compelling. At a minimum, if you're serious about calibration which I think you would be otherwise you wouldn't be on these threads, try the comparisons for yourself. Theory only goes so far, then you have to do your own testing.

(Flamesuit on) smile.gif

I agree, after reading this thread the only definitive answer for a particular Plasma (in the case of the OP its a 65VT60) is there is non. Then again that has been the jest of other threads that have tried to answer the window pattern question for Plasma's.
As you said Jim "you must try for yourself".

ss

Note; I am not saying the theory's, explanations and specs or the others before this thread are not helpful because they are to lead me to this conclusion. smile.gif
post #56 of 193
Quote:
Originally Posted by LarryInRI View Post

Of particular interest was the data and discussion concerning the chromaticity shift in the 20% to 40% stimulus range when comparing standard windows on a black background and windows on a constant APL background. I think this finding deserves more consideration.

Larry

yes, people tend to think that the ABL circuit is the only thing that modifies the transfer function but it's not even active until average relative luminance is greater than about 25% if you have peak white set to modest values, say 120 cd/m^2 (meaning most of the time ABL is not modifying the luminance for movies). So the gamma and chromaticity shifts that Tom and I found in that stimulus range (and the one you measured) are probably due to some non-linearity of the plasma drive at very low average luminance.
post #57 of 193
Quote:
Originally Posted by Doug Blackburn View Post

Still baloney. Constant APL patterns exercise ONESINGLE point in the plasma ABL spectrum. You cannot stop ABL when viewing images on plasmas, but constant APL windows also produce a calibration at a single/constant ABL which will never happen when you view content on a plasma. Your calibration should include at least SOME range of ABL action on the part of the plasma TV and the only way to do that is with traditional window patterns with a black background.

Avoiding critical evaluations of high APL images against the LCD "reference"... OK, in a practical sense you can't do that comparison. But that implies being rather cavalier about the appearance of images when the plasma's ABL is getting more significant... if you only care how the low end of the luminance range looks, (as that's what you are comparing to the LCD), how will you ever know if your constant APL calibration patterns are doing justice to the plasma calibration when ABL begins to be more of a factor? You won't. Which is why you really need to have SOME ABL action in your calibration. I'm no advocating intentionally pumping up the ABL, but I am advocating not eliminating it entirely from the calibration (as you do when using constant APL patterns).

I have NO IDEA what you mean by "large patterns" If you mean "full screen" or anything larger than 10%-12% of the physical size of the screen, that's just a red herring, nobody (I know) would advocate calibrating a plasma display with full screen or even 20% or larger window patterns. If you are talking about 5% physical windows as being "proper" and 10% windows as being "large".... I suggest using something more descriptive. Like "10%=size".

And by making 1 comparison, you can't make any sweeping conclusions because, as mentioned in the previous post, different brands and models of plasma have different APL "action" so what you perceive as working well for 1 particular plasma, could be a bad choice for some other plasma.

Very interesting all that your commenting.

So you suggest that because of the constant changes by APL it would be best not to calibrate the gamma and trusting that engineers who designed the screen do their job properly?

In other words. Better NOT touch gamma in palsmas?

I understand that due to this feature of plasma technology is impossible to know with certainty what the correct image composition

Excuse my english
post #58 of 193
Quote:
Originally Posted by Doug Blackburn View Post

Constant APL patterns exercise ONESINGLE point in the plasma ABL spectrum.

I think the patterns JimP mentioned do not actually hold a constant APL during the measurement series, since "10, 18, 25% surrounds" may indicate that the APL changes by the area of the measurement window. For example a 5% area window on a constant background varies APL by 5%, between the pattern with a black measurement and one with white. If the background is black the APL changes from 0%-5%, and if the background is 22% then APL varies approximately 21%-24%.

I think some of the measurements on this site indicate it's possible to measure the same video level with a similar APL pattern and get a different measurement on some plasmas. Getting a non-matching measurement from identical video levels and APLs might suggest that the display reacts to average luminance of the pattern, which seems to be what zoyd was suggesting. The patterns that JimP and LarryInRI mention all appear to vary average luminance during the measurement series. I think the patterns LarryInRI was thinking of using on the new "calibration" vary 0%-6.5% average luminance for windows, and somewhere in the range of 4%-11% or 3%-13% average luminance for the "APL" pattern, assuming a 2.2 gamma. Personally I don't know why someone would ideally want to measure gamma using patterns that vary average luminance if the display has the ability to vary luminance for the same picture level (video level) depending on changes in average luminance.

Quote:
Originally Posted by Miguelronda View Post

So you suggest that because of the constant changes by APL it would be best not to calibrate the gamma and trusting that engineers who designed the screen do their job properly?

Far as I can tell, he's saying that windows with a black background are magic, but maybe there is actual measurement data somewhere to support the opinion.

Quote:
In other words. Better NOT touch gamma in palsmas?

I'm not sure this conclusion is in the comments, yet I think it makes more sense to just leave 10 point gamma controls alone if you have an unverifiable reference. Between the side-by-side reference against a non-adjusting display like JimP is suggesting, or windows with a black background, personally I don't know how the window option is necessarily a better automatic choice.

Quote:
Originally Posted by LarryInRI View Post

Of particular interest was the data and discussion concerning the chromaticity shift in the 20% to 40% stimulus range when comparing standard windows on a black background and windows on a constant APL background. I think this finding deserves more consideration.

I'm not familiar with the pattern you used, but if the pattern does intend to hold APL constant using two video levels, the calculated minimum average luminance falls in this range. I'm not suggesting the dip in average luminance is necessarily the cause of the increased difference in blue from the RGB balance measurements, since that sort of comment would require more measurements.
Edited by alluringreality - 8/21/13 at 1:39pm
post #59 of 193
I've posted all I'm going to on this topic. You should try it for yourself and draw your own conclusiions.
post #60 of 193
Quote:
Originally Posted by alluringreality View Post


Far as I can tell, he's saying that windows with a black background are magic, but maybe there is actual measurement data somewhere to support the opinion.

There is none, the argument stems mainly from historical inertia without any metrology basis. I summarize here my findings in a little easier to understand format regarding why one shouldn't calibrate plasmas with any pattern than generates less than ~5% average relative luminance[ARL] (to use Poynton's terminology).

These graphs were generated using 1% windows on fixed level backgrounds to produce the series of luminance levels plotted. At each level a 20 point grey scale run was performed to calculate gamma and dE. The background stimulus levels used were 0%, 2%, 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, and 50%. [240 total measurements plus stability checks]





The chromaticity shift is relative to 10% windows on a 22% background and I've plotted the worst of the levels (30%). Using this dE scale it's equivalent to ~2 JNDs. This is in agreement with previous findings by both Tom H. and I that chromaticity shifts are less of an issue than gamma instability. However, both average gamma and the 30% level shifts in particular are significant in my opinion. That along with the fact that 99% of anything you will ever view will exceed 5% average luminance there is a good argument for avoiding windows on black backgrounds. There is another subtle effect regarding color saturation that occurs when using standard windows. Because they measure lower for gamma, the display will be adjusted to darken those levels and this in turn pushes colors towards the gamut edge. Not by much but it is measurable, about 1.5 dE94 for the red channel on my display.

note also that peak white was stable to within 2% at ~140 cd/m^2 for all runs, so the general perception that ABL plays a roll in these variations is false. In fact ABL plays little or no role in most of the video images you will encounter. Broadcast video and movies have average relative luminance predominantly in the range of the "grey world" assumption 15%-18% or lower. Plasma ABL circuits typically cut-in above 25% ARL, which makes perfect sense from an engineering design perspective.


Edited by zoyd - 8/22/13 at 4:47pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Display Calibration
AVS › AVS Forum › Display Devices › Display Calibration › Gamma with normal windows or with APL?