UP2720Q, software issues with color calibration

Thank you for staying connected. I'm Vivek, and I will be assisting you.

If I have understood correctly, you’re facing an issue with the Monitor Calibration. We will look into this for you right away. I will need your help in performing some simple steps to identify what’s causing this.

Could you please share a picture of both the screen so that we can have it reviewed?

Hi Vivek,

Thanks for having a look at my issues. I attached 2 pictures to illustrate my first issue. Each pictures shows a part of my first UP2720Q, while I pasted a sample of the second UP2720Q screen in the middle of that picture to clearly show the differences. 

The top picture is the factory sRGB preset (all is good here). The bottom picture after I did a recalibration of that preset. The built-in colorimeter should improve the factory calibration, but it does not. But monitors measure worse after recalibrating (verified with Xrite I1 pro) and, they both measure different from each other (both wrong, delta E2000s over 5).

I would also appreciate it if you could have a look at my other, software related questions. Maybe they go hand-in-hand.

BR!

Factory calibrationrecalibrated using internal colorimeter

We request you to private message all of this for secure conversation.

PM sent.

1- Your i1ProX spectrophotometer cannot measure it properly using Xrite or Calman software. At 10nm readings you'll face about 3dE error if that UP has some kind of GB-LED variant. If it uses some WLED PFS variant (which is extremely likely) you won't be able to measure it properly at all with that software, it cannot see red channel spectralpower distribution. IDNK which one is used in this UP, older 2016 UPs where WLED PFS variants.
If you validate with ArgyllCMS/DisplayCAL and its 3.3nm mode for i1ProX you'll measure it OK if it was a GB-LED but still will be some little error if it was a WLED PFS variant but you can see actual SPD ant 3nm.
i1ProX devices are not really menat to measure displays, at best to act as helpers to Xrite colorimeters using 3party softwae like DisplayCAL. These new "narrow spike spectra" backlights liek WLED PFS or laser projectors are really out of scope of your device using Xrite or CALMAN based software.
Also its low light readings are not good.
Hence: unless you used DisplayCAL 3nm high res mode your claims about factory calibration on spot may not be true at all.

2- Check manually for profiles in "C:\Windows\System32\spool\drivers\color", order by date reverse. Then go to classic control panel, color management, device tab, use this config for my device, add that profile... but it should be there.
If not found because one of the many bugs in Dell HW cal software, just profile without GPU calibration using DisplayCAL+ArgyllCMS. As I said previously your device is not able to measure properly some backlights and even using ArgyllCMS' 3nm mode poor low light readings will be there.
While i1displaypro is a HID device, like a mouse, so it doe snot need drivers, an i1ProX needs an USB driver for using it with DisplayCAL. It is not compatible with Xrite based apps, so you'll be forces to switch between drivers when changing from ArgyllCMS based apps to Xrite/CALMAN based apps:

http://argyllcms.com/doc/Installing_MSWindows.html

Very easy to do driver switch, no reboot, just a few clicks... but you have to do it.

Once USB driver is installed, move to your HW calibrated preset, change all "calibration tab" settings to native in DisplayCAL (= no calibration), choose profile type and create profile.
Try first a simple matrix single curve profile with black point compensation. Your device is far to slow to try something else at 1st try.
At the end you get a ICM profile, you can use DisplayCAL tray loader for GPU calibration. Since it contains no calibration (GPU LUT input=output) it won't hurt. Also if you own an AMD card you can do GPU calibrations this way without banding (thanks to DisplayCAL loader and AMD HW+SW)... you cannot do the same using windows LUT loader, it causes truncation (although AMD dither may cover it partialy).

Finally (Your "finally" quote) I did not know that it had such limitations regarding HW cal software, I thought it was DUCCS compatible. Thanks for the information. I agree with you that is a big limitation.
Consider returing them and buy 2 x Eizo CS2740. About same price. It is a 27" UHD display. HW calibration is provided through Color Navigator 7. Good support with an i1DisplayPro. With an i1ProX ypu'l face the same innacuracies because 10nm mode in some new backlights.

Also buy an i1DisplayPro colorimeter, really.
i1ProX 10nm measurements are not meant for this new backlights. Also speed constratins will hold you back from making very datailed profiles for using them as inputs for LUT3D creation (Resolve or home content with madVR). i1ProX is just a helper regarding displays, but a very useful tool when making printer profilers or measuring fabrics, worth every penny you paid for it in these tasks... but no so good with displays.
Using your i1ProX with ArgyllCMS at 3nm you can correct every possible colorimeter for backlights and even use Argyll  made 3x3 matrix correction (at 3nm resolution) with CALMAN software (since calman itselft is not able to do it that way with that precision).

Note i1Prox = i1Pro, i1Pro2. Model i1Pro3 is not supported by ArgyllCMS and AFAIK it won't be supported soon.

Hi @yumichan,

Happy to see your reply here.

1. The panel definitely has a PFS-based red phosphor (lots of narrow spikes). Validation was done using Argyll with -H parameter, so 3.3nm resolution. I agree with you on the small error, but did not calculate the resulting dE variance.

2. I actually found out that Dell Color Management Software was the culprit. I downloaded the first version released and that fixed the ICC problem. Profiles are being generated and saved just fine now. Your Argyll-method was indeed my last resort solution as well.

3. I cannot seem to get a decent answer form Dell regarding current or future DUCCS compatibility. I can open and run DUCCS, but cannot see the CAL1 CAL2 tables, so it seems to resort to GPU calibration. Not sure if this is normal behavior. 

Another issue: Dell Germany does not sell CalMAN for Dell. This means I can never take advantage of the advanced features of the monitor!

I am honestly not very convinced. High price, no support, completely broken or non-existing software and...maybe the most important issue: a very narrow viewing angle. I have distinct pink-brown color shifts when I view things off-axis. Maybe this is typical (?) for "newer" 4K wide-gamut IPS panels, but my older ones do not have this at all.

2 of the Eizos you mentioned would cost me 700 Euros more in total...not sure if I can afford that right now. The I1's main task is profiling WF printers here, and up to now I was very happy with its consistent calibration of different monitors. 

Sigh...

DUCCS seems to be on dead end. i1Profiler moved to 3.x to support new i1pro3 and some corrections and we still use some 1.7.x Custom i1Profiler for these dells. Other brands updated their failty HW software for this new tools (Benq)... but no news from Dell. This calman stuff looks like defunction note for DUCCS.
Little hopes...

Also there is no update for Dell SDK, latest 2.0.0. Check it but I think that your UP2720Q is newer and not supported. It is possible to code in C++ a program to upload LUT-Matrix-LUT calibration to these monitors with that SDK.
If D65 native gamut is good for you, LUT-matrix-LUT contents can be simplified to:
preLUT = linear
matrix = identity
postLUT = .CAL contents from a profile made with displaycal at uncalibrated CALx, calibration in GOU to whateever white (D65 in this sample), then interpolated to cover 0-1023 entries.
(pre & post LUT should contain degamma and re encode gamma, but with an identiy matrix for gamut (native) you can do this "linear lut" trick )

That C++ program is simple no special maths involved, just use DLL communicate with monitor and upload data in lut-matrix-lut format described in SDK doc. Data can be interpolated even with excel...

Unfortunately my U2413 is not supported by that public SDK, so I cannot test it... because they do not want to support it (DUCCS does the same task on my U2413, so there is a library even if they do not want to publish it for my model)

If your monitor can use that SDK you can avoid that useless calman stuff.