[argyllcms] Re: Noticeable differences between calibrated white points
- From: Sam Berry <samkberry@xxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Tue, 26 Oct 2010 11:26:08 +0100
On 26 October 2010 00:23, János, Tóth F. <janos666@xxxxxxxxxx> wrote:
> Well. This is not a part of my job, I only do this calibration thing as a
> hobby (a supplementary for my hobby, like home cinema). So, my goal is the
> adequate viewing experience, not some kind of hard correspondence for
> standards.
>
> But I have two displays right now. I placed them next to each others (I
> power up the secondary when I need more place for my CAD drawings and other
> documents at the same time --- for example, a standard in A4 paper format
> and a Mathcad window to see what should I draw...).
> The main display is a H-IPS and the secondary is an S-PVA. The first one is
> a 10 bit display but the VGA card is able to do some dithering on the LUT
> output, so I can calibrate it with effectively 10 bit precision. I think it
> should be enough for "something".
> The instrument is a ColorMunki and I use it in Adaptive HiRes mode (but I
> tried the normal HiRes mode as well - it kills the shadow detail but I want
> to talk about something else here...) and I use the Black/White measurement
> drift features during the High quality calibration (in ArgyllCMS/dispcal, of
> course :)
>
> They always showed different colors because the H-IPS has a very wide gamut
> (but I mostly use it in hardware emulated sRGB mode) and the S-PVA has a
> near-AdobeRGB gamut. So, it's obvious that they should show different colors
> when I don't do any gamut conversion with softwares (and the purpose of the
> secondary display is mostly black and white document viewing, so...) and
> even if I do, the S-PVA can't cover the full sRGB or AdobeRGB gamuts while
> the H-IPS has a ~100% coverage. So, I never expected that they will show the
> same colors.
>
> BUT I expected that they will show the same white after a fresh
> calibration. But no...
> I watched black and white documents on both of them at the same time and I
> noticed that they show very different whites. So, I thought I should
> recalibrate them. But it didn't help.
>
I run a dual montior setup. I don't have a CCFL H-IPS, but I do have an
S-PVA, an LED H-IPS and a CCFL S-IPS, and to be honest, I;ve found that the
only way I can get the H-IPS to match any of the others is by eye,
spectrometer or not. All the others will match pretty well but the LED IPS
eludes me, by about 6dE 2k.
>
> The measured dE2k error was 0.3 and 0.0 dE2k after the calibrations. My
> instrument used to be in the dE2k 0.3 range between continuous measurement,
> so the highest possible dE2k should be under 0.9 (at max ; relatively, I
> don't speak about the absolute error...).
> But they shows noticeably different whites!
> I also tried to measure these numbers with a "cross-acclimatized"
> instrument (I re-checked the first display after I finished the calibration
> on the second display). There were no any significant difference (it was a
> luck but it read the exact same dE=0.0 on the H-IPS).
>
>
> I invoked an EIZO document where they stated that you need FOV10 observer
> for color matching displays.
> OK, I used the FOV10 observer (and the correct x;y coordinates but it
> doesn't really matter when we talk about relative differences).
> The only difference that both displays have a reddish tint now. And the
> relative difference is equal, or may be higher (but not smaller at all).
> Now what?
>
It should be noted that wide gamut displays look significantly different
between the two observers, and so how you compare the two screens becomes
very important. Filling the two screens with white and looking at them will
result in a white which does not match when comparing small patches of
colour. Similarly, matching the two screens with a small, central monochrome
image with good gradation (preferably of something nearly monochrome in real
life to aid perception) with a black surround give the best results for me,
although the two screens look miles off when looking casually. My LED screen
looks great calibrated with the i1 to D65, but when looking at small
details, it becomes obvious that they have a green cast. It's a strange
effect. My S-PVA wide gamut has about half this effect, and my s-IPS and CRT
almost none.
Calibrating with a 10 degree observer will help match casually, but won't
help you judge colour better, in my experience.
>
> The instrument usually does a good job until you don't have two displays
> next to each others to judge. If the uncalibrated display feels
> bluish/reddish/greenish, then the calibrated state will feel as "white".
>
>
> Is it normal? Or is my instrument broken or is it unable to calibrate the
> WCG display, or what...?
>
Unfortunately eyes are not as consistent as we would sometimes like to
think, never mind the instruments we make to emulate them. I set my D65 on
my normal-gamut S-IPS and match anything else to that by eye. I also find
that using the profile made from the visually wrong but colorimetrically
correct white point generally works better than one done after correcting
the white point, as the extra chromatic adaptation performed is usually
unwanted.
>
>
> I don't know if it can make a significant difference in the perception of
> the white color when it's filtered and mixed from different spectrums. Is
> it? (Sorry, if this whole email was a noob question.)
>
>
> I made these pictures with a smartphone, so you should ignore those
> horizontal and vertical lines. I wanted to catch the relative differences,
> not judge about "which one is whiter":
>
I would ignore any pictures taken to compare white points, if you don't
trust your spectro, it's very unlikely that your camera phone is more
colorimetrically correct :)
The calculator thing is indeed a polarisation issue, and my H-IPS is
polarised 90 degrees away from most other LCDs. Try turning the calculator
sideways :)
Sam Berry
www.satsumatree.co.uk
Other related posts:
