or Connect
AVS › AVS Forum › Display Devices › Display Calibration › Display Calibration Calculator
New Posts  All Forums:Forum Nav:

Display Calibration Calculator - Page 5

Quote:
Originally Posted by thomasl

Tom, I think this also happens when you have an undersaturated but bright color - at least according to my brief experiments. CIELUV reports a lower dE with the higher brightness compensating for the undersaturation whereas CIE94 reports a dE that is worse. In my mind, it seems to boil down to whether you think brightness changes in a color help to offset errors in saturation. CIELUV says 'definitely yes' while CIE94 mainly says 'not really'.

Exactly. That puts it well.

I hadn't considered whether the 1976 formulas would also report color error attenuation when an undersaturated color is made excessively bright, but you right to say that they do.

However, the effect is not nearly as strong as when oversaturated colors are made excessively dim. So for the oversaturated green 0.296, 0.696, 0.715, CIELUV reports that the maximum amount by which you can lower the dE simply by lowering the brightness (to 0.495) is from 28.2 to 13.3 (52.8%). For a similarly undersaturated green of 0.304, 0.523, 0.715 CIELUV reports that the maximum amount by which you can lower the dE simply by raising the brightness (to 1.01) is from 28.2 to 17.9 (36.5%). BTW, CIELAB reports that even larger changes in brightness are required to achieve maximum reductions in dE. So this is a feature of the 1976 formula. It has nothing to do with the Lab vs. Luv color spaces.

CIE94 reports nothing of the sort. In fact, accoring to this metric these changes in brightness actually raise the dE somewhat rather than substantially reducing it.

AVS Top Picks

Some people in another thread were bothered that 0.1-0.2 dE differences (that are invisible to the eye) for primary and complementary colors were being calculated by some CIELUV spreadsheets and my Display Calibration Calculator. The reason for the differences was that my calculator used only 3 decimal place precision for the standard (Rec 709, SMPTE C, EBU) yellow, cyan, and magenta x,y values when calculating dE, and at least one spreadsheet used 4 decimal place precision for those x,y values, which created differences.

I increased the internal calculation of those standard yellow, cyan, and magenta colors to 16 decimal places !!! That should make everybody happy now. (Other internal calculations were already 16 decimal places.) The new PC and Mac versions (v1.24) are now posted at www.accupel.com under HDG-4000 Manuals on the left side of the page. (That's either to encourage you to download the HDG-4000 manual and decide that you can't live without one - or I was too lazy to create a separate page for the calculators - you decide.)

You may still see 0.1 dE round-off differences between my calculator and some spreadsheets that use only 4 decimal places for the x,y complementary color targets. I increased the precision to 5 decimal places in one spreadsheet and then it agreed with my calculator in every example I tried. But there is still the possibility of a visually invisible (that sounds so cool ) 0.1 dE round-off error unless everyone uses 16 decimal places like the Display Calibration Calculator does.
Of course, Excel has issues with rounding at the 14th decimal place, so let the support questions roll!
Yummy, a MAC version.
Quote:
Originally Posted by mlaun

Yummy, a MAC version.

There is always a Mac version when I do something. Feel lucky if you get a PC version too!
too bad you don't do calman I run it with bootcamp. Runs well in vmware as well... but native would be real cool. Sorry was dreaming!
Quote:
Originally Posted by gregr

Some people in another thread were bothered that 0.1-0.2 dE differences (that are invisible to the eye) for primary and complementary colors were being calculated by some CIELUV spreadsheets and my Display Calibration Calculator. The reason for the differences was that my calculator used only 3 decimal place precision for the standard (Rec 709, SMPTE C, EBU) yellow, cyan, and magenta x,y values when calculating dE, and at least one spreadsheet used 4 decimal place precision for those x,y values, which created differences.

I increased the internal calculation of those standard yellow, cyan, and magenta colors to 16 decimal places !!! That should make everybody happy now. (Other internal calculations were already 16 decimal places.) The new PC and Mac versions (v1.24) are now posted at www.accupel.com under HDG-4000 Manuals on the left side of the page. (That's either to encourage you to download the HDG-4000 manual and decide that you can't live without one - or I was too lazy to create a separate page for the calculators - you decide.)

You may still see 0.1 dE round-off differences between my calculator and some spreadsheets that use only 4 decimal places for the x,y complementary color targets. I increased the precision to 5 decimal places in one spreadsheet and then it agreed with my calculator in every example I tried. But there is still the possibility of a visually invisible (that sounds so cool ) 0.1 dE round-off error unless everyone uses 16 decimal places like the Display Calibration Calculator does.

Greg, something is wrong with the dE calcs for secondaries in the new calculator. They are way too high. See attachment
Quote:
Originally Posted by kjgarrison

Greg, something is wrong with the dE calcs for secondaries in the new calculator. They are way too high. See attachment

Yeah, I see it. I introduced a bug when I made the quick change to increase the precision. I checked the previous version and it works correctly. Looks like if you set the Y target value to 1.0 (top set of numbers) then the current version works correctly. Give me a few minutes to find the bug and then I'll create new versions and load them to the accupel website. I'll post here when they are fixed.
I just posted new PC and Mac versions of the Display Calibration Calculator. I introduced a bug in the last version when I increased the precision of the internal calculations. (YIKES! How embarrassing!) The dE calculation for Yellow, Cyan, and Magenta was totally wrong if the target value for Y was anything other than 1.00. In this version (1.25) I also increased the number of displayed digits for Yellow, Cyan, and Magenta to 4 decimal places so that dE will equal zero if you exactly copy the x,y values from the targets to the measured values.

Thanks to kjgarrison for finding and alerting me to the bug. Please let me know if I messed up anything else when I created version 1.24 (or the quick fix version 1.25). Sorry for the error.
Greg, I have a question about the behavior of your calculator vs "others" out there.

With the Accupel calculator, using the measured data which is not on spec for Rec709 and trying to find the best target Y values as determined by minimum dE for each color, each change in Y for a given color only effects the dE value for that color.

"Other calculators" behave differently. Some result in a change of all dE values indicating a different target for all Y values if any one changes. My understanding is that they are following the rule that the sum of all primary colors' Ys should equal white Y.

Is this rule only applicable to Rec709 (or perhaps any Rec) situations, and not applicable to non-Rec "real world" situations? Or is this too simplistic of a question?

Here's a thread where Bill is explaining his calculator methodology.
http://www.avsforum.com/avs-vb/showt...4#post16401154
Quote:
Originally Posted by kjgarrison

Greg, I have a question about the behavior of your calculator vs "others" out there.

With the Accupel calculator, using the measured data which is not on spec for Rec709 and trying to find the best target Y values as determined by minimum dE for each color, each change in Y for a given color only effects the dE value for that color.

That is correct. If you change the measured Y value for red it will obviously change the dE for red (dE simply compares the measured red color to some target red color). Changing the measured Y value of red doesn't change the measured Y value of another color, hence it doesn't affect the dE for other colors.

Quote:
Originally Posted by kjgarrison

"Other calculators" behave differently. Some result in a change of all dE values indicating a different target for all Y values if any one changes.

As you say, if all dE values are changing when the measured values of one color are changing, then the target colors must be changing. But the target colors normally come from the standards (Rec 709, SMPTE C, EBU) so you don't want the target colors to change as you make CMS adjustments to try to achieve the target colors. That makes no sense. The CMS objective is to hit the target colors for minimum dE, not to the change the targets as you make adjustments. If you need to adjust the Y value of a color differently than the specified Y value of its target color in order to minimize the dE of that color (because if has different x,y values than its target color) that is absolutely no reason to then adjust the x,y,Y values of the other target colors, which is apparently what is being done. That makes no sense physically or logically.

Quote:
Originally Posted by kjgarrison

My understanding is that they are following the rule that the sum of all primary colors' Ys should equal white Y. Is this rule only applicable to Rec709 (or perhaps any Rec) situations, and not applicable to non-Rec "real world" situations? Or is this too simplistic of a question?

The concept that the sum of the primary colors Y value should equal the white Y value (or the sum of two adjacent primaries Y values equals the Y value of the complementary color) is only valid when there is no 6-axis CMS. i.e. it is true when you check/adjust color decoding with NATIVE primaries, but not when adjusting a 6-axis CMS. Hence, the AccuPel calculator allows you to copy the measured NATIVE primary x,y values and the x,y,Y value of the measured white back into the target section and then compute the primary Y and complementary x,y,Y values for the purpose of adjusting the color decoding, not for the purpose of adjusting a CMS. When adjusting a 6-axis CMS, summing the Y primary values to equal the white Y value, or summing primary Y values to equal a complementary Y value, no longer has any physical or logical relevance. It is akin to the infamous "straight line" issue we disposed of earlier. (Technically speaking, those relationships hold for a linear system, but a 6-axis CMS is a non-linear system composed of six linear (usually - but they don't have to be linear) color gamut segments. Although linear relationships exist within the linear color gamut segments they do NOT exist across segments, i.e. when involving more than one primary or complementary color.

All of this confusion about "straight lines" and "adding Y values" apparently comes from not understanding that a 6-axis CMS is a non-linear system (usually made up of 6 linear parts). The linear relationships of a 3 primary system do not apply across the 6 color gamut segments. When we discuss the "linearity" of a 6-axis CMS (which you will see discussed in other threads), we are talking about linearity within the 6 color gamut segments, i.e. how well do the x,y,Y values for a particular primary or complementary color track with incoming signal levels. Only if the CMS is perfectly linear within each segment, AND if each segment is perfectly aligned with a linear color gamut target (i.e. Rec 709), does the resulting non-linear CMS then produce an effectively linear color gamut mapping from the native primaries to the target (i.e. Rec 709). In that case, and only in that case, do the linear relationships apply across the color gamut segments, and that is the only time we want them to apply. Otherwise, trying to force these relationships actually produces larger errors from the target, because that is the same thing as trying to achieve a target based on other primaries. (Perhaps this is getting too technical, but you can't change the Y value of a target primary without either changing the target white point or changing the x,y values of the target primary or target primaries.)

Glenn

Edit: Never mind. Found it thanks to a fellow member.
I'm looking for the calibration calculator, as well. I originally downloaded the Mac version a couple of years ago and recently tried to find the PC version, but I can't seem to find it anywhere. I'd appreciate if anyone could point me in the right direction.

Thanks!
I am also looking for this calculator to download if anyone knows where to find it...thanks....
It looks like the calculator that Greg had originally on accupel.com has moved or been removed. I think he is working with Tom Huffman now. I'm not sure of any of the details but here is the link to the calculators that Tom has on his website. They are very good! http://www.chromapure.com/colorscience-calculators.asp
New Posts  All Forums:Forum Nav:
Return Home
Back to Forum: Display Calibration

AVS Top Picks

AVS › AVS Forum › Display Devices › Display Calibration › Display Calibration Calculator