Wide Gamut monitor calibration: i1 Display Pro better than DTP94?

[jcs1]

Having read that modern colorimeters such as the i1 Display Pro were more suitable for calibrating wide gamut monitors I bought an i1 Display Pro and compared the RGB readings with those of the DTP94 using ColorNavigator 6.3.0.11.

The monitor is a ColorEdge CG243W. The target is: gamut monitor native, brightness 90 cd/m2, black level 0.4 cd/m2, white point 6000K, gamma 2.20, priority grey balance.

The RGB results after adjusting the monitor to this target are presented here as the mean values of 5 “Adjustments” using DTP94  and 4 “Adjustments” using  i1Display Pro:
R (0.6769, 0.3114 DTP94) vs. (0.6732, 0.3141 i1DPro)
G (0.1904, 0.6981 DTP94) vs. (0.1991, 0.6975 i1DPro)
B (0.1521, 0.0585 DTP94) vs. (0.1525, 0.0574 i1DPro).

I have not done any statistics but variance within each mean value is low and the differences therefore look genuine. It is my understanding that the numbers show that i1Display Pro calibrates to less saturated R,G and B than DTP94.

The resulting White Point is 6004K for DTP94 and 5994K for i1Display Pro. A grey calibration card looks slightly redder after calibration with the latter.
Delta E values (maximum, mean, white point) are similar for both colorimeters when the monitor is “validated”.

Could someone kindly explain to me what this all means besides variability between colorimeters? Can we conclude from these results that i1Pro is a more appropriate colorimeter than DTP94 for the Eizo CG243W? If not, which observations led to the view that i1Display Pro is better than DTP94 for calibrating wide gamut monitors?

Thank you.

Jean-Claude

[deejjjaaaa]

If not, which observations led to the view that i1Display Pro is better than DTP94 for calibrating wide gamut monitors?

I guess you read the article written quite quite some time ago by Dry Creek Photo - that was the observation I think.

[D Fosse]

If you have an i1 Display Pro, there is no reason to even consider using anything else. It is widely known as a brilliant piece of equipment for all modern displays.

(I assume the Dry Creek piece is this one: http://www.drycreekphoto.com/Learn/Calibration/MonitorCalibrationHardware.html

Quote: "The DTP-94 excels with standard–gamut displays but fares poorly on wide-gamut models").

[deejjjaaaa]

If you have an i1 Display Pro, there is no reason to even consider using anything else.

with $$$ I'd buy BasicColor device... but alas

[D Fosse]

with $$$ I'd buy BasicColor device... but alas

Well, yes...but at $1300 I might decide to spend that money elsewhere. The i1 D3 is good enough for all practical purposes.

There will always be small differences between sensors, but mostly within the larger variable of finding the exact right calibration targets. Which can be tricky enough. I use an Eizo CG246, which comes with its own internal sensor (a supposedly high-end Konica/Minolta sensor), but I also have the i1 D3. They're different, but which one is "best" is almost a moot question. They just require very slightly different target settings to arrive at the same result.

[GWGill]

Could someone kindly explain to me what this all means besides variability between colorimeters? Can we conclude from these results that i1Pro is a more appropriate colorimeter than DTP94 for the Eizo CG243W? If not, which observations led to the view that i1Display Pro is better than DTP94 for calibrating wide gamut monitors?

No, you can't conclude much in a technical sense from any of these observations.

To determine which of the two instruments is more accurate, you need to compare both to a reference instrument, i.e. something like a CS-2000 Spectroradiometer.

[Czornyj]

with $$$ I'd buy BasicColor device... but alas

Not a big difference, probably slightly better inter instrument agreement, maybe slightly better sensitivity in bkpt measurement, but i1D3 is also very good in this regard. My impression was that the DISCUS relies on precise calibration to panel+backlight spectrum to a larger degree than i1D3, so it sucked each time such calibration wasn't perfect. i1D3 seems to be more forgiving in this regard, I suspect it has filter characteristic that's closer to standard CMF.

IMO i1D3 is good enough - it's not perfect, but once you start to freak out in this regard, you'll soon end up with Spectrascan PR-740, Konica-Minolta CS-2000, or at least JETI Specbos 1211 + Klein K-10.

[deejjjaaaa]

you'll soon end up with Spectrascan PR-740, Konica-Minolta CS-2000, or at least JETI Specbos 1211 + Klein K-10.

everybody is entitled to his/her own sex... gear fantasies

[deejjjaaaa]

and some off topic... which monitors with h/w lut capabilities inside allow to use 3rd party (any for that matter) software to write to those LUTs... NECs/Eizos do allow at least a BasicColor to do this surely, Dells do not (only their own OEM software) - what about others ? HPs, BenQs, etc, etc ?

[Czornyj]

None (AFAIK). Dell, HP, LG use OEM/X-Rite based solutions.

[jcs1]

Thank you all for your replies from which I have learned a lot. They made me take a second look at my observations and extend my comparison between readings of DTP-94 and i1Display Pro with a slightly altered white point target of the latter.

The observed trend for the DTP-94 to show more saturated R, G, and B colour points than the i1Display was not confirmed. Due to the variance between successive calibrations more than the 6 additional calibrations I carried out are required to confirm or disprove this trend with reasonable certainty; I feel the relevance of the question is not worth the effort. All readings of both colorimeters lie at the limit or slightly outside of the Adobe RGB gamut and are very close to each other.

In the excellent cited paper by Ethan Hansen (Dry Creek Photo) the accuracy of several colorimeters and spectrophotometers used for monitor calibration was tested by comparing their metering of a defined white point and a defined black point with those of a high-end spectrophotometer (PR-730). The mean of these readings made with the DTP-94 deviated from the PR-730 readings to a considerably larger extent than the readings of the i1Display Pro. The accuracy of metering individual colours was not reported.

Based on these data and the practical experience reported by those who replied to my questions it seems that replacing the DTP-94 with the i1Display Pro for the calibration of my GC243W is a good idea after all. Thank you.

Jean-Claude
 

[Czornyj]

In the excellent cited paper by Ethan Hansen (Dry Creek Photo) the accuracy of several colorimeters and spectrophotometers used for monitor calibration was tested by comparing their metering of a defined white point and a defined black point with those of a high-end spectrophotometer (PR-730). The mean of these readings made with the DTP-94 deviated from the PR-730 readings to a considerably larger extent than the readings of the i1Display Pro. The accuracy of metering individual colours was not reported.

Based on these data and the practical experience reported by those who replied to my questions it seems that replacing the DTP-94 with the i1Display Pro for the calibration of my GC243W is a good idea after all. Thank you.

The wtpt deviation from reference is a sum of R,G,B errors, as white is 100% of RGB channels.

As you can see in Ethan's evaluation, DTP-94 has good inter instrumental agreement, and EIZO ColorNavigator has correction matrices for supported colorimeters, so in practice the deviation from reference should be smaller than in the above mentioned test.

[jcs1]

The wtpt deviation from reference is a sum of R,G,B errors, as white is 100% of RGB channels.

As you can see in Ethan's evaluation, DTP-94 has good inter instrumental agreement, and EIZO ColorNavigator has correction matrices for supported colorimeters, so in practice the deviation from reference should be smaller than in the above mentioned test.

“My reply to “The wtpt deviation from reference…”
 In my first post I commented regarding Ethan’s paper “The accuracy of metering individual colours was not reported” as I wondered whether it is correct to assume that if a colorimeter is accurate for the white point it is also accurate for (non-white) colours. Is it possible that a colorimeter’s accuracy for certain colour patches that are sent to the screen (e.g. by ColorNavigator) is less satisfactory than for the white point (and hence should be evaluated when comparing colorimeter accuracy)? If the answer is “no” I wonder why we use ColorChecker Passport or similar tools to assure colour accuracy of photographs and not just a grey card? Which colour patches are used to test white point accuracy: only 255,255,255 or others as well? I would be grateful if someone could clarify these questions for me.

My reply to “As you can see in Ethan's evaluation…”
I presume you refer to the deviation in the data I reported in my first post. Thank you for letting me know this. What can I do to make the deviation from reference smaller? The monitor and the colorimeters were warmed up, I calibrated in the dark and all calibrations and validations were done in the same (rather long) session. Could this be caused by a problem with the graphics card? I previously noted that I got better dEs with the DTP-94 after changing (for the same Eizo CG243W) from a 17” MacBook Pro5,2 to a 2009 Mac Pro4,1 (which I presume has a better graphics card, i.e. a NVIDIA GeForce  GT 120).

Thank you.

Jean-Claude

[Czornyj]

“My reply to “The wtpt deviation from reference…”
 In my first post I commented regarding Ethan’s paper “The accuracy of metering individual colours was not reported” as I wondered whether it is correct to assume that if a colorimeter is accurate for the white point it is also accurate for (non-white) colours. Is it possible that a colorimeter’s accuracy for certain colour patches that are sent to the screen (e.g. by ColorNavigator) is less satisfactory than for the white point (and hence should be evaluated when comparing colorimeter accuracy)? If the answer is “no” I wonder why we use ColorChecker Passport or similar tools to assure colour accuracy of photographs and not just a grey card? Which colour patches are used to test white point accuracy: only 255,255,255 or others as well? I would be grateful if someone could clarify these questions for me.

If the display is well linearised a simple ICC profile with correct colorant chromatic coordinates can characterise it well. We use some validation color patches to evaluate how good is the characterisation is, and if display gamut is large enough to reproduce certain colors.

I presume you refer to the deviation in the data I reported in my first post. Thank you for letting me know this. What can I do to make the deviation from reference smaller? The monitor and the colorimeters were warmed up, I calibrated in the dark and all calibrations and validations were done in the same (rather long) session. Could this be caused by a problem with the graphics card? I previously noted that I got better dEs with the DTP-94 after changing (for the same Eizo CG243W) from a 17” MacBook Pro5,2 to a 2009 Mac Pro4,1 (which I presume has a better graphics card, i.e. a NVIDIA GeForce  GT 120).

Graphic card is irrelevant to validation results, as long as the connection is digital (DVI, DP). Don't worry too much about it, it only shows the difference between profiling measurement and validation measurement, it's rather to evaluate if the display and sensor was stable, and if profiling measurement went right (eg. wasn't interrupted by screensaver etc.).

[jcs1]

Thank you Marcin  for this useful information.

Jean-Claude