- This topic has 92 replies, 3 voices, and was last updated 3 years, 10 months ago by
.
-
Posts
-
Seems that my monitor BENQ EW3270U is affected by metamerism/observer failure as the D65 whitepoint done with the 1nm, CCSS sample found in the database looks way to blueish compared to my standard IPS monitors which closely matches white on my professionally calibrated OLED with alternative whitepoint.
OLEDs with alternative WPs are RGB OLED not consumer WOLED TVs. Consumer WOLED TVs with “numerical” alternative WP to D65 are just poorly measured TVs in most cases. (Note the difference between “numerical” vs “visual/perceptual” where that consumer TV is visually matched to another display that is its reference)
“standard IPS” monitors unless calibrated to D65 with an accurate device… are not D65. “standard IPS” monitors whitepoint is unknown unless measured.
There is/was a JETI CCSS for that displays, almost a perfect match to “Panasonic VVX**** ccss” (95% P3), so that CCSS works unless manufactured changed backlight model.
The whole setting you try to compare is hardly consistent so aiming to “numerically” alternative WP is an error unless you:
-Measure all of them choosing the best colorimeter correction for all or a reference device.
or
-use visual white point match for all, choosing one as “reference”P.S: On top of that, of course there is the possibility of observer metameric failure… but that is user related and varies with user so the whole purpose of alternative “numeric” white points on a WLED PFS monitor is a nonsense. It has to be a “visual” match.
About the only way to fix such a metameric failure is to use either an accurate wide filter spectrally calibrated colorimeter (e.g. i3 Display Pro or Spyder 5) or preferably use a spectrophotometer.
Given such calibration argyllcms will calculate the nearest metamers and place them in LUT. You might have to use a very big LUT though as interpolation will produce wrong results.About the only way to fix such a metameric failure is to use either an accurate wide filter spectrally calibrated colorimeter (e.g. i3 Display Pro or Spyder 5) or preferably use a spectrophotometer.<br>
Given such calibration argyllcms will calculate the nearest metamers and place them in LUT. You might have to use a very big LUT though as interpolation will produce wrong results.This was done with a i1d3 and a 1nm CCSS file the EW3270U measured around D65, seems like Vincent is right that only a visual match would be correct
Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Speaking of white point, DisplayCAL reports 4 values. I know what CCT is, but what is the VCT calculated from?
(There are two versions for daylight and blackbody locus.Mind you, you skipped the *spectrally calibrated* part. Just using i1d3 without the right correction will produce wrong results, different corrections will cause variation of up to 1000K in white point. You need a spectrophotometer to calibrate the colorimeter for this specific type of screen. At least the results are reasonably portable between various instances of i1d3.
I mean, the CCSS is probably correct for the display that was measured. Not necessarily *your* display. If it’s a sharp CCSS, and it is, a simple 10 nm offset within calibration will ruin matches. And that’s definitely within manufacturing tolerance of the LEDs.
I’m having such an issue with Acer XV273K here, which seems to be PFS type phosphor but not matching any of them exactly. There is a nice matrix profile made by someone, but it produces nonsense numbers, for example VCT of ~6500 K for uncalibrated state (though ~5700 K CCT) where it actually looks pinkish, kind of like whiter Illuminant E. (And has delta E of ~16.)
Speak of “alternative” white point. It is getting corrected to slightly bluer version due to calibration mismatch, reported as 7000K VCT, but 6500 K CCT. (All temps daylight locus.) On gamut chart it looks dead on, but it’s slightly wrong.I mean, the CCSS is probably correct for the display that was measured. Not necessarily *your* display. If it’s a sharp CCSS, and it is, a simple 10 nm offset within calibration will ruin matches. And that’s definitely within manufacturing tolerance of the LEDs.
I’m having such an issue with Acer XV273K here, which seems to be PFS type phosphor but not matching any of them exactly. There is a nice matrix profile made by someone, but it produces nonsense numbers, for example VCT of ~6500 K for uncalibrated state (though ~5700 K CCT) where it actually looks pinkish, kind of like whiter Illuminant E. (And has delta E of ~16.)<br>
Speak of “alternative” white point. It is getting corrected to slightly bluer version due to calibration mismatch, reported as 7000K VCT, but 6500 K CCT. (All temps daylight locus.) On gamut chart it looks dead on, but it’s slightly wrong.The CCSS file is for the same display model though, and it’s a 1nm spectral sample, I would expect the whitepoint to look the same as a standard gamut LCD, even the OLED TV I have looks the same when at D65, it’s basically only this monitor that it seems I need to calibrate to around 6000k VDT to make it match D65 of those two others,. If someone else only had this monitor then they would have no way of knowing that the measured D65 looks way too blue compared normal gamut ones.
Same model is not good enough for sharp peaks like OLED and PFS phosphors, with LED manufacturing tolerances being what they are sometimes. If you used a spectrophotometer it’d just work with the correct number. It’s exactly the same problem I have with my display – the primaries will look to the colorimeter as if they are offset a lot, but they actually are just minimally missing the calibration. It will look on the resulting chart vs reference like they’re moved. If you “calibrate” with this wrong setup, argyll will chop off a hunk of gamut. (Here, it chopped off about 3% of DCI-P3 needlessly.)
I’m literally buying i1 Studio to verify this theory. (I’m almost certain that’s it. In my case, red missing the calibration done by some internet person by a few nm.)
The idea here is, even if i1 Studio does not have the proper accuracy, it will likely find the mean offset from the 1nm calibration you have, which you can then modify by hand using DisplayCAL Python APIs.
The idea here is, even if i1 Studio does not have the proper accuracy, it will likely find the mean offset from the 1nm calibration you have, which you can then modify by hand using DisplayCAL Python APIs.
Idk the PFS phosphor one found from xrite almost matches exactly the 1nm CCSS sample found in the database, my own CCSS file done with i1 2 pro with 3nm driver also is very close to the 1nm one.
Then there should really be no problem with calibration. There’s really no possibility of a metameric failure here with that wide filters i1d3 has… unless it’s sending near UV and you’re picking it up. Which is rather impossible. And the filters are stable over time.
What’s the uniformity of the screen? Perhaps you’re measuring in an off color patch?
You can always dump the observer curves of your i1d3 from Argyll with an ifdef. Someone should turn that into a command line option, as it’s useful if you bought a recalibrated one from ChromaPure.
Then there should really be no problem with calibration. There’s really no possibility of a metameric failure here with that wide filters i1d3 has… unless it’s sending near UV and you’re picking it up. Which is rather impossible. And the filters are stable over time.
What’s the uniformity of the screen? Perhaps you’re measuring in an off color patch?
You can always dump the observer curves of your i1d3 from Argyll with an ifdef. Someone should turn that into a command line option, as it’s useful if you bought a recalibrated one from ChromaPure.I dunno man all 3 monitors use correct CCSS profiles and are very close to D65, the EW3270U one still looks way too blueish at D65, when I perceptually matched to the other two that at least looks the same at D65 the EW3270U was missing green mostly needing to change from R50 G47 B50 to R50 G50 B43, one of the displays that I matched to was even done by a professional using 1nm spectros (LG C9) ,my i1d3 sample manufactured in 2019 with correct CCSS was measuring pretty close to the guys jeti being only 20K off which is not noticeable.
It can’t be that 3 different CCSS files are wrong when they measure almost exactly the same?
I doubt it’s uniformity as I can’t see differences anywhere in color temperature when displaying 100% white in full screen,and the measurement was done in the middle which is pretty uniform when moving the i1d3 probe.
Interesting. How far are the primaries from the target color space after calibration?
Color can only look blue (and still report 6.5k VCT) if x coordinate is off by negative offset. According to certain report (linked on AVSforum), i1d3 can have problems sometimes in x plane with accuracy out of the box, reporting false smaller x values. Recalibrating it fixes that. (I1 Studio having different offset accuracy issue, a cone of confusion mostly due to observer positioning, producing warmer results rather than offset.)
Same model is not good enough for sharp peaks like OLED and PFS phosphors, with LED manufacturing tolerances being what they are sometimes.
No. RGB OLED peaks are not narrower than a QLED or gree-blue or a GB-LED or blue from a common sRGB WLED.
They claim that RGB OLED blue go to shorter wavelengths in the left downhill and there CIE 1931 2degree may not match a significative amount of population.Then there should really be no problem with calibration. There’s really no possibility of a metameric failure here with that wide filters i1d3 has
Again… no. You get all in a wrong way. Metameric failure is not because i1d3 filters. *If firmware curves match actual i1d3 curves* whatever mismatch vs some standard reference observer it is corrected by an accurate CCCS. Metameric failure because observer happens because of YOU, or ME, or they. XXX standard observer may not really match a certain individual “observer”
Hence => Visual white point editor as A.ces or myself said before.An i1pro or i1pro2 or i1studio are unlikely to get you a better CCSS than 1nm sample from community sample made with a JETI (EW3270U) or Xrite default sample for 95% P3 WLED PFS (and they should match => excel & plot them + RGB “gain” match)…. *unless manufacturer chose to change backlight to other cheaper P3 type without notification (like those “nano” from LG or some WLED PFS with variation in floor in red channel shorter wavelengths as you may see by plotting Xrite PFS family collection vs Panasonic VVX CCSSs)*.
Then instead of using a cheap spectro as reference, plotting each display WRGB SPD made with that cheap spectro side by side may shed light on this: actual change in backlight type or flavor = it WON’T be a manufacturer “tolerance” issue within some panel part, but an actual deliberate backlight type change (or change from some WLED PFS to another WLED PFS ***flavor***)If your i1d3 curves do not match firmware curves you’ll need old way colorimeter to spectro correction, a matrix. 3nm argyll driver high res if you cant get a better device. 1st match numerically, then from that starting point, visual whitepoint if they do not match.
