Problem with absolute XYZ values measurement (X-rite Colormunki)

[Jan Roik]

Hi,
Most recently I tried to create a correction matrix for the X-rite colormunki probe. For this purpose, I used results measured by this probe and the results obtained by the spectrometer. Although the relative values of xyz were the same the absolute XYZ was not. XYZ values measured by the probe were like 36x bigger. This wasn’t really a problem because I just multiplied XYZ values obtained from spectrometer by constant and then I successfully created a calibration matrix but I wonder what is the source of this difference.
Any ideas in this regard?
Moreover when I tried to measure illuminance in ambient mode with (colormunki) “spotread” it returns Y= illuminance [lux]. On the other hand, when I calculated illuminance from the original XYZ obtained by spectrometer the results differ dramatically from this obtained by the probe.
It is not essential to obtain the same XYZ values but, I should be able to obtain the same illuminance from these devices.
Can I trust measured illuminance by colormunki?
One more question.
Is there any software correction between measurement in an ambient and emissive regime?

Thanks for any suggestion.

[Vincent]

Hi,
Most recently I tried to create a correction matrix for the X-rite colormunki probe. For this purpose, I used results measured by this probe and the results obtained by the spectrometer. Although the relative values of xyz were the same the absolute XYZ was not. XYZ values measured by the probe were like 36x bigger. This wasn’t really a problem because I just multiplied XYZ values obtained from spectrometer by constant and then I successfully created a calibration matrix but I wonder what is the source of this difference.
Any ideas in this regard?

Use builtin matrix correction creation in DisplayCAL instead of manually compute a FCCM. Or write how did you measure reference parches which each device, that could reveal if you did something wrong.

One more question.
Is there any software correction between measurement in an ambient and emissive regime?

Thanks for any suggestion.

If you mean colormunki display, or any other i1d3 colorimeter, emissive & ambient modes store its own set of different spectral sensivities.
If you make a CCSS correction from an ambient source (with spectro in ambient mode) you should be able to feed it to ArgyllCMS.
To port a CCMX in a mode to the other mode will be extremely difficult without some knowledge about actual spectral distribution of the source (because emmissive and ambient modes store different spectral sensivity curves in firmware, they are different)… so you depend on a CCSS again.

Calibrite Display SL on Amazon  
Disclosure: As an Amazon Associate I earn from qualifying purchases.

[Jan Roik]

I used ccxxmake -f to crate CCMX correction. Ref.ti3 was measured by spectrometer and Targ.ti3 by colormunki Display. The problem was the difference between the absolute values of XYZ of those two measurements. The difference between these values caused that diagonal elements of the CCMX correction matrix were too big (around 4). I just multiply my ReF.ti3 data with an average difference between XYZ values (x36). This led to CCMX correction with diagonal elements (around 1). I assume this result is correct. But I am curious why there is a difference between XYZ values in the first place. For example, XYZ of the white color measured by Colormunki Display (92.56 100.00 115.95) and by Spectrometer (2.86 2.99 3.98). I wonder if it is only coincidence that Y=100 for the white color or if it is intention and white is standardized and base on that XYZ values of other colors are measured.

So argyll just read values from the probe both in ambient and emissive mod and don’t apply any correction? In the case of the ambient mode when Spodread return lux values this is also read from the probe or this is calculated by Argyl? Because when you calculate illuminance manually you use formula Y*682 if your color matching functions are defined for the maximum of Y=1. But illuminance returned by Spodread is equal to Y value. Does this mean that color matching functions, in this case, are defined for a maximum of Y=682?

[Vincent]

For example, XYZ of the white color measured by Colormunki Display (92.56 100.00 115.95)

It looks like a normalized value, I didn’t test to create ambient matrix correction but if it fails/it has a bug, try to use spotread and get actual Y value, then modify “normalized” colormunki display value to actual CIE XYZ values.

[Florian Höch]

What are we measuring? A display? Argyll’s dispread in its current version does not offer ambient mode readings. Furthermore, the use of ambient mode for display readings only potentially makes sense for non-contact measurements, e.g. projector off the lens instead of off the screen. For instruments that have an ambient mode, selecting this mode usually selects a special ambient calibration matrix (or SPD) inside the instrument. There is no additional software processing for ambient readings.

color measured by Colormunki Display (92.56 100.00 115.95) and by Spectrometer (2.86 2.99 3.98)

How did you obtain these values, exactly? As Vincent mentioned, the CM Display readings seem to be normalized (Y=100), and this is always the case for files created by dispread (the actual absolute luminance is in the LUMINANCE_XYZ_CDM2 field). unless the -w flag is used (not recommended).

[Jan Roik]

There are actually two measurements. First is on display, in this case, I used emissive mode and “dispread” function. I measured a small sample of colors generated by DisplayCAL. If I understand it correctly, in this case, the first color in the sample is white so “dispread” always set Y=100 and then other colors are measured base on that? This explains my problem with the difference between absolute XYZ values between probe and spectrometer.

In the second measurement, I measured 3 bulbs. In this case, I used ambient mode and function “spotread”. This function returns illuminance in [lux]. I am not sure about the value obtained by the probe. Because from the previous measurement I found out that the obtained results are the same up to constant. So I see no reason why I cannot measure the same illuminance with these devices. Why is illuminance equal to Y value in the case of the Colorminki Display? Is there a need for some previous calibration when I want to measure in ambient mode or I fust plug the Colormunki Display into USB and start measurement?

[Jan Roik]

I found the same old X-rite I1 and I tried the same ambient mode measurement and “spotread” function and illuminances I have obtained were the same as values measured by spectrometer. So there is probably a problem with Colorminki display probe.

[Florian Höch]

If I understand it correctly, in this case, the first color in the sample is white so “dispread” always set Y=100

See above.

Why is illuminance equal to Y value in the case of the Colorminki Display?

When using ambient mode, Y reported by spotread is in lux for any instrument.

I can’t take ambient measurements with my i1 Pro because I miss the ambient cap, but if I compare emissive measurements of ambient daylight reflected off the ceiling walls, using spotread, to those of my i1 Display Pro (roughly same spot and position), then they mostly agree (within about 7%) despite this being an old i1 Pro Rev. A, which is known to report slightly too low luminance values (compared to a newer device like an i1 Pro 2).

So there is probably a problem with Colorminki display probe.

I think that’s unlikely.

[Jan Roik]

Ok, thank you for all your replies. I really appreciate your help. It helped me a lot.

[Author]

Posts