Huge difference in measured vs assumed white point

[Moortgat Leo]

I’ve been using displaycal quite some time now with a spyder express.

This week I bought an xrite i1.

I understand most of the settings and how they affect the results.

But last nigth I was running a calibration on my laptop screen. The measured uncalibrated white point was about 7200K.

The settings for white point were set to ‘as measured’ and so I calibrated the screen.

Afterwards I ran a measurement report, without any verification profile .

All the results were OK , except for the white point. The delta was about 9 points of , saying assumed vs measured white point mismatch.

But when I looked at the report, it said about 7200K and then 6500K daylight.

Since I used ‘as measured’ i’m a bit confused.

So I did the calibration again, and now I put 6500K instead of as measured.

Guessing this would be even worse, in the final report.

But what I didn’t expect , was that this verification report was OK for white point.

Since it’s a laptop screen I can’t adjust RGB levels, so the white point is always fixed to about 7200K.
I did try before with Intel Graphics Panel, adjusting the contrast of the RGB levels, and that seemd to work to reduce the white point to 6600K.

But yesterday I read on this forum that you shouldn’t do it this way. And leave the white point ‘as measured’.

And that brings us back to the start of my questions 😉 .

Some extra info, I’m running win10 , I use a second display Benq with 100% srgb for my LightRoom/PhotoShop work.
That runs fine calibrated. It’s just the laptop screen that has me ‘puzzeled’ and moreover how to interprete the white point settings.

Thanks in advance.

[Vincent]

But last nigth I was running a calibration on my laptop screen. The measured uncalibrated white point was about 7200K.

The settings for white point were set to ‘as measured’ and so I calibrated the screen.

So no calibration of white. Report chose the closest daylight white to native white. White color (excluding brightness) is 2D position in a map, correlated color temp says nothing about actual color (missing info). Assumed vs measured shows color distance between current white and “closest daylight white” in 100K interval.
Since no calibration => huge errors in white are common.

Since it’s a laptop screen I can’t adjust RGB levels, so the white point is always fixed to about 7200K.
I did try before with Intel Graphics Panel, adjusting the contrast of the RGB levels, and that seemd to work to reduce the white point to 6600K.

Do not use control panel. If you choose D65 as target (6500 AND daylight, 2 crossing lines in that 2D map) for white, displayCAL will limit some channels output in a better way that what you can do on control panel. All these translate to GPU LUT correction in VCGT tag in resulting ICC profile.
Do nothing on control panel, just set a whitepoint target if you do not want native white.

For visual matching your other benq use visual whitepoint editor as whitepoint target in DisplayCAL.

[Moortgat Leo]

Vincent,

That makes sense, I guess. I never had this problem with the Benq because it could set the RGB levels to reach 6500K (almost).

So it’s best to calibrate the white point to 6500K and choose 6500K over ‘as measured’. Hence it’ll skip white point calibration.

One other question. How usefull is it to use a lux reading of the ambient light?
Assuming I always do my edits during the same conditions?

Thanks in advance.

And another thanks for pointing me to the xrite instead of the spyder5, it’s way faster and more accurate!

[Vincent]

Native = as measured = max contrast, max unique grey levels

The further you go from there, the more contrast you loose. It may be negible if distance between native & target is small.
Choose the leser evil. Another option between them is choosing the closest daylight white as target, it may be cool (bluish) but white (no green or pink).

Since some TN laptops have very reduced contrast as measured is default recommended value. As long as you know what are you doing and why… choose the one that best fuit your needs. For example IMHO is better to avoid greenish bluish cast on white on some office laptops even if you loose contrast but YMMV.

[Moortgat Leo]

Vincent,

Ok. So if I leave it at ‘as measured’ for the TN panel. It’ll give me a warning in the report.

But all the other values are verry good.

So the only downside is that my native white will be bluish?

[Vincent]

And grey and all the surrounding colors.

Verification “all green” just means “display and profile match” / “profile predicts display behavior”.
It does not mean “it works as I expect it to work”.
It could mean “if i do not care about overall XXXX-ish cast on screen, color managed apps can work OK and show accurate content on a relative whitepoint intent”.

[Moortgat Leo]

Today I noticed that the displaycal profile loader, reapplied the calibration state +500 times!!

And this on a timespan of 3 hours.

Is this normal?

[Vincent]

Yes, it just ensures that no other app like some games cleans GPU LUTs.

[Author]

Posts