- This topic has 25 replies, 4 voices, and was last updated 2 years, 9 months ago by
.
-
Posts
-
Dear All,
im new and not super experienced with display calibration, just a bit.
Im using several displays for working on photographs. Sometimes a quato monitor (IP 242 LE), sometimes only the notebooks display and sometimes an Eizo CG2730.
One question: does DisplayCAL Profil Loader and Eizos Color Navigator work together when the notebook is on the Eizo Monitor? Or do the work against eachother somehow?
How to handle it with my several displays?
Thanks a lot for helping here!
Greetings!
Timo
Eizo agent will try to load into GPUT LUTs the VCGT data from ICC profiles (HW cal profiles will have linear data = no GPU calibration). They may conflict.
Since users do not use ColroNavigator so often, it’s easier to disable CN agent auto startup & user DisplayCAL loader. When you need to recalibrate or change CG HW calibration from another one stored in your computer, close DisplayCAL tray app, open CN, do what you need, after that unload CN agent and reload DisplayCAL loader.
I’ve recently calibrated Eizo CS2731, the first Eizo that I’ve met since I’ve become friendly with ICM. I used Color Navigator only, though DisplayCAL was installed on computer. There’s no conflict between profile switching with Eizo display menu and DisplayCAL Profile Loader. Moreover, Profile Loader is useful to control current ICC profile, that Eizo device sets (using USB connection) as Windows display default. But, there’re two troubles:
- Switching Eizo display profiles (for different calibrations) in display menu is buggy: sometimes it doesn’t change Windows display profile. But if you navigate through calibrations menu backward, this restores. This doesn’t look like conflict with Profile Loader, but Eizo software/firmware bug only.
- Eizo display showed some wrong whitepoint after calibration. Its incomfortability was sensed by eyes too. I think the best way here is to use a bit shifted xy coordinates instead of color temperature. I used this approach today, but for wrong color cast got by DisplayCAL at laptop. I find that using DisplayCAL over Color Navigator (to get better profile) may produce problems, at least if you use quick profile switch in Eizo menu. Another way to get better white point is using User mode in Eizo. You can calibrate it too, but brightness and RGB bars are unlocked here, so you may do slight correction of white point. This won’t make significant influence on overall tolerance.
- Eizo display showed some wrong whitepoint after calibration. Its incomfortability was sensed by eyes too. I think the best way here is to use a bit shifted xy coordinates instead of color temperature. I used this approach today, but for wrong color cast got by DisplayCAL at laptop. I find that using DisplayCAL over Color Navigator (to get better profile) may produce problems, at least if you use quick profile switch in Eizo menu. Another way to get better white point is using User mode in Eizo. You can calibrate it too, but brightness and RGB bars are unlocked here, so you may do slight correction of white point. This won’t make significant influence on overall tolerance.
If you are not using latest CN, update. If using 7.1.2 Change colorimeter to no correction or compensation, I do not rememmber (preferences/general) then retry. Since is a GB-LED try to compare in DisplayCAL with the same EDR/CCSS (RG phosphor, a “dirtier” GB-LED U2413 with other displays).
If this does not work then you have to use visual witepoint in CN (but this goes to HW not to GPU like in DisplayCAL)
@Алексей Коробов,
Did changing colorimeter settings to “no compensation” solveyour issues with CN & CS GB-LEDs?
Did you check by P3 coverage is they are using GB-LED type in newer CS or moved to WLED PFS?I haven’t tried anything yet, the display is now 100km away of my house, this is my friend (a girl) who bought it. But probably I’ll pass her i1d3 for a week, so we’ll check it. This display doesn’t have full P3 coverage, it has around 95%. It is quite better than BenQ PV270 in yellow, but quite worse in cyan (press colors, – I has Epson 9900 with CMYK+OGV 3rd party inks filled in, still fixing it and profiling it with RIP). Prad.de has tested it and it looks like truth with my friend’s instance. Unpleasant thing was a little “crumpled rag” view of white area.
I think now that my i1d3 is not perfectly tuned for normative RGB values, – likely to be tired. So I’ve made several attempts with matrix correction between i1p2 and 1id3, and this works better. I also prefer to set WP xy to 0.3123 0.3180, that make the procedure more stable to pink cast and makes a bit yellowish one. I note also that my second-hand i1p2 seems to have a bit degraded white tile (though it looks well , i1p2 was sold to me by local serviceman, I know him as a diligent man). I also think that Argyll could use some dirrerent values or measurement calculations than i1Profiler or Color Navigator. Not a serious difference, of course.
Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.I also think that Argyll could use some dirrerent values or measurement calculations than i1Profiler or Color Navigator. Not a serious difference, of course.
It takes the same readings if you choose in Argyll the same correction as the other apps. With i1Profiler it’s easy: use the same names. All EDR corrections in i1Profiler are translated to CCSS in i1d3 bundle for displaycal (actually Displaycal has more): https://displaycal.net/i1d3
AFAIK, for CN if you use “No compensation” on CS line it will take only RG_phosphor EDR, so readings in Argyll with i1Profiler’s RG_phosphor (U2413) CCSS should be the same.
With other than no compensation setting I’m afraid they are using a matrix correction from a CS sample in factory but since no i1d3 is exactly equal that matrix is not valid for all. I say that I’m afraid of that because Spectraview II for WLED PFS AdobeRGB + P3 models (PA271Q + PA311D) uses that approach, a chain of corrections, 1st GB-LED (PA242W EDR, which contains no colorimeter data, it’s an EDR) and chained to it, it applies a matrix (“one sample unit” colorimeter specific) hence all the issues with i1d3 and those models (mva.pl polish forum, LuLa… etc)
It’s absurd this approach of matrices instead of bundle HP Z24x WLED PFS AdobeRGB+P3 EDR… but maybe there are some royalties or IP involved.Also I’m afraid that since CN stores no EDR for WLED PFS AdobeRGB + P3 models (it will be PFS_Phosphor_Family_31Jan17 -dirty- or HP_DreamColor_Z24x_NewPanel EDR, the good one), just RG_phosphor it will have the same issues on CG models… BUT with “no compensation” you can detach factory colorimeter matrix correction and use EDR correction in ColorNavigator (NEC can’t) so in theory if you replace spectral samples in RG phosphor file with HP z24x EDR spectral data (repeated several times, or even just replacing files)… voilà! 😀 CN + “no compensation” + “forged EDR” = accurate CN readings on newer Eizo CGs and an i1d3.
That could be apply for Benq SW owners too since they lack of proper correction for all models …BUT RGBLED EDR spectral samples have different nm sizes (HP Z24 and GB-LED have the same nm range in samples). It may need some hex editor work if “file replacement & renaming” does not work (if you have to edit manually data in RGBLED EDR to “forge” it)
Since there are no Xrite corrections for QLEDs like SW2700PT, this tool can be helpful:
https://github.com/ypomortsev/ccss2edr (it’s not mine)
Although if file replacement is not valid you may need to interpolate QLED CCSS to 1nm first (and even scale data to RGBLED nm range), convert 1nm CCSS to EDR, open it, copy spectral data and replace in RGBLED EDR, then run Palette Master Elements & calibrate. Grey innacuracies will be there but white readings should be on spot on every i1d3 (calibration may be off due to oversimplifications in PME programming).Eizo display showed some wrong whitepoint after calibration. Its incomfortability was sensed by eyes too. I think the best way here is to use a bit shifted xy coordinates instead of color temperature. I used this approach today, but for wrong color cast got by DisplayCAL at laptop. I find that using DisplayCAL over Color Navigator (to get better profile) may produce problems, at least if you use quick profile switch in Eizo menu. Another way to get better white point is using User mode in Eizo. You can calibrate it too, but brightness and RGB bars are unlocked here, so you may do slight correction of white point. This won’t make significant influence on overall tolerance.
Hey Alexey! Did you managed to solve this CS2731 whitepoint problem?
My search for solution: https://hub.displaycal.net/forums/topic/confusion-calibrating-cs2731/#post-30759
Hope you have some ideas 🙂
I haven’t met any Eizo since that times. But now I use i1p2 and i1d3 both for calibration/profiling with DisplayCAL in most cases, and the key step is personal matrix correction for my i1d3. Yes, these devices aren’t fine, non-personal spectral correction produces inacceptable results in many cases. But… for WLED displays (that’s not your case) there is white perceptance problem too. i1p2 may show ideal WP, but you see pink or yellow cast. This is model problem that hasn’t resolved yet in DisplayCAL. I’m still thinking on “by eye” white choice methods (using reference white card and high-CRI light, by example).
If CN allows to measure custom white, I recomend you to get its coordinates from calibrated display of similar type, like Macbook Pro. This must be done in CN, not in DisplayCAL, cause it uses different colorimeter correction. All in all evaluate resulting WP by your eye. Actually our eyes adapt our vision to different WPs, so slight difference is not the case, relative color distances is much more valuable factor.
i1p2 may show ideal WP, but you see pink or yellow cast. This is model problem that hasn’t resolved yet in DisplayCAL.
If it is a WLED PFS as user made correction suggest (the other thread from Midas)… your i1p2 cannot measure it with extreme accuracy even at 3nm.
DisplayCAL cannot correct spectrophotometer limitations, Argyll custom driver can take readings from sensors without averaging to 10nm like Xritte driver.. but cannot go beyond that because there are no more sensors to read at shorter wavelength jumps.
The only way to correct it is to buy or rent a JETI or CR and do not use i1p2.Also it it was an observer metameric failure issue, DisplayCAL cannot correct it, No program cancorrect that. Just provide some visual WP adjustment, more or less user friendly.
Anyway, at 3nm it should be good enough, at least to make a CCSS and feed it to a i1d3 relying on factory spectral sensivities:
I didn’t do the maths, but if somebody is willing you’ll have to take HP Z24x 1nm reading, perform some averaging to transate it to 3.3nm series, then integrate both spectral power distributions with CIE 2 degree observer (1931 or 2012) and see how color coordinates drift with bigger wavelength jumps. That will be a “hint” of actual error of a generic but accurate 3nm Xrite spectro measuring WLED PFS with high res driver. My guess is that it is small.
Same exercise at 10nm (Xrite, Calman, basiccolor, lightspace, Color Navigator, Spectraview II, Dell & Benq software…), my guess is that it will be big.If CN allows to measure custom white, I recomend you to get its coordinates from calibrated display of similar type, like Macbook Pro.
Macbooks are not and cannot be used as reference because they are not properly calibrated to D65 CIE 1931 2degre. Grey calibration is fine but white is not D65 Several samples in this forum. Well, white looks white on most of them (<3dE to daylight locus at cooler CDTs than D65) so thay can be used as everyday computer display without noticing.
Yesterday I made calibration of the same Eizo CS2731. Now it looks pretty good. I have also made some tests to investigate problems shown above and in parallel branch (https://hub.displaycal.net/forums/topic/confusion-calibrating-cs2731/). I did not made profile with DisplayCAL, but with CN only. My conclusion is described below.
- The last version of Eizo Color Navigator (CN) does very good job if you use i1 (i1p2 in my case).
- I’ve found that profile change bug (made from display menu) is fixed. You can also assign custom names for display menu entries (not CAL01 etc.).
- DisplayCAL tests with i1p2 and i1d3 (+ mtx personal correction) both show a bit cooler WP, around +50K, I shifted WP in CN to 0.3132 0.3293 later, this gave 6504K at the output (CN always adds 0.0001 to target xy for unknown reason, you see it in test).
- i1p2 and i1d3 + mtx personal correction, based on that i1p2, show some different contrast (~1000 for i1p2 vs. ~850 for i1d3). That is strange, other values in test reports are close to each other. Contrast depends on gamut used (830:1 by i1p2 for sRGB + evening-tuned WP). Sometimes I meet this effect on other panels. I haven’t tested profiles with CN yesterday cause I considered DisplayCAL as testing platform.
- My i1d3 being used in CN with embedded correction makes significant magenta cast. I haven’t measured this cast, but it looks like dE=4…7 to D65. Display native WP is close to this cast. Probably CN embeds correction for standard phosphors, but G or R has some shift cause of manufacturing change or aging. Here I note that my BenQ PV270 factory gamut (official ICC gamut) is close enough to Eizo CS2731 factory gamut, but BenQ panel uses different phosphor for R (hill-like curve against double-peak of Eizo).
- CN uses binary correction files, we don’t know how to decode them, so we can’t use DisplayCAL/Argyll-built i1d3 personal correction for it.
- I used iterative WP xy coordinates look up (to get almost perfect WP D65): I switched i1p2 driver to Argyll after closing CN and made test, then I predicted better xy for new CN calibration, etc. I recommend this way for i1d3 owners, while they can make good personal mtx correction using my spectral data (ZIP attached). I also downloaded spectral correction (and my own i1d3 mtx correction) to online database (taged with “by Korobov” ).
- You can’t set xy coordinates in User1 mode, but brightness and other tuning bars are unlocked. Here I corrected WP with RGB bars and DisplayCAL pre-calibration dialogue. This may be less precise, but I used this mode for “consumer display” emulation (sRGB gamut + gamma 2.2).
- I forgot to check grey gradient for new profiles, but CN has two options for purifying it, I hope it is has no problems.
- ColorMunki Design showed significantly different test result, WP especially. I haven’t used it hereafter.
- Uniformity test was made with i1d3 (DUE display correction mode was set to uniformity), max dE=1.49 at the left found, dE=1.2 found at the top display area. More than expected.
Uniformity test is attached.
UPD: I missed CN EDR correction hack in parallel branch. But is HP Z24x spectrum really equal to Eizo CS2731? They have different panels. Does anybody can compare my data with 1nm-stepped HP data?
Yesterday I made calibration of the same Eizo CS2731. Now it looks pretty good. I have also made some tests to investigate problems shown above and in parallel branch (https://hub.displaycal.net/forums/topic/confusion-calibrating-cs2731/). I did not made profile with DisplayCAL, but with CN only. My conclusion is described below.
- The last version of Eizo Color Navigator (CN) does very good job if you use i1 (i1p2 in my case).
10nm… if it is a WLED PFS details missing on reading
- My i1d3 being used in CN with embedded correction makes significant magenta cast. I haven’t measured this cast, but it looks like dE=4…7 to D65. Display native WP is close to this cast. Probably CN embeds correction for standard phosphors, but G or R has some shift cause of manufacturing change or aging. Here I note that my BenQ PV270 factory gamut (official ICC gamut) is close enough to Eizo CS2731 factory gamut, but BenQ panel uses different phosphor for R (hill-like curve against double-peak of Eizo).
Which one? MTX or EDR. MTX (defauult) is done for a generic i1d3 that may not match yours. EDR is RG_phosphor EDR. If you need to use other spectral correction check Midas’ thread. Inclusing some hints about EDR format.
Just repeat 5 times the 4 WRGB spectral samples of what you consider “reference”: Since you have an i1Pro2 measure backlight in displaycal at 3nm and see which LED it is useing. Then you can choose the closest generic from other EDR or interpolate to 1nm your i1Pro2 readings.
- CN uses binary correction files, we don’t know how to decode them, so we can’t use DisplayCAL/Argyll-built i1d3 personal correction for it.
Explained a little on Midas thread. if you want to use custom CCSS made with your i1Pro2:
-Create CCSS 3nm
-Interpolate to 1nm using whatever software is easier to you
-CCSS2EDR to pack text CCSS to float EDR
-Replace 5×4 samples in RG_phosphor EDR with your new Reference (5 displays identical to your CCSS) with an HEX editorOR
(it seems that Midas used this)
Replace 5×4 samples in RG_phosphor CCSS in display CAL with your manually interpolated to 1nm CCSS made with i1Pro2 (3nm)
Then use CCSS2EDR to repack it into a RG_phosphor EDRThen replace original EDR in CN (close app first, replace, run etc).
- I used iterative WP xy coordinates look up (to get almost perfect WP D65): I switched i1p2 driver to Argyll after closing CN and made test, then I predicted better xy for new CN calibration, etc. I recommend this way for i1d3 owners, while they can make good personal mtx correction using my spectral data (ZIP attached). I also downloaded spectral correction (and my own i1d3 mtx correction) to online database (taged with “by Korobov” ).
AFAIK there is no easy way to use your own personal MTX (instead of your custom CCSS) because it must be hardcoded on some DLL or EXE. Very difficult to locate & patch.
if you use a MTX created with 3nm reding and a CCSS made that way on most i1d3 there should be very little dfference between them, so use the “forge EDR approach” as described above to make your custom CCSS into ColrNavigator… but since you want your own reference (instead of HP Z24x G2 EDR/CCSS), you have more work to do: interpolate 3nm to 1nm.
EDIT: check that starting nm and ending nm match the original RG_phosphor EDR. If not you must add or remove spectral samples till it matches exactly what RG_phosphor stores:
SPECTRAL_BANDS “351”
SPECTRAL_START_NM “380.000000”
SPECTRAL_END_NM “730.000000”- Uniformity test was made with i1d3 (DUE display correction mode was set to uniformity), max dE=1.49 at the left found, dE=1.2 found at the top display area. More than expected.
Not really, dE=delta color + delta brightness. In the same report choose deltaC, it should be very low (mostly +-X brightness error without color tint).
For most of us we can live with some small delta L, but want very low delta C. <2dC should be no color tint.Uniformity test is attached.
UPD: I missed CN EDR correction hack in parallel branch. But is HP Z24x spectrum really equal to Eizo CS2731? They have different panels. Does anybody can compare my data with 1nm-stepped HP data?
Backlight = LED technology, not panel. According to these 2 users:
https://colorimetercorrections.displaycal.net/?get&type=ccss&manufacturer_id=ENC&display=CS2731&instrument=i1%20DisplayPro%2C%20ColorMunki%20Display%2C%20Spyder4&html=1It is a WLED PFS, AdobeRGB green flavor => like HP z24x and current CG in Eizo , but measure with your i1Pro2 at 3nm
Since EDR can have some intellectual property rights it seems not wise to attach the forged EDR here, but maybe he can send you using fichier or MEGA. Otherwise, HEX WORK.
