- This topic has 88 replies, 13 voices, and was last updated 3 months, 2 weeks ago by
.
-
Posts
-
Visually I get purplish looking blues when I calibrate with 2012 and i1 Pro colorimeter.
This is with a colorimeter profiled against a Jeti 1501 2nm hi res.
Colorimeter is useless since it has CIE 1931 filters. Entire calibration needs to be done with spectro alone if you want to target alternate CMF’s.
Then it is not properly profiled (CCMX), or the “cloud of points” cannot be fitted to that idealization.
Colormunki Display doesn’t have filters for other CMF’s. You need a colorimeter that is compliant for CIE2012.
You need a spectro if you want to calibrate in other CMF’s.
As far as I understand the ColorMunki Display can load colorimeter correction files for the specific display (CCSS,CCMX) which allows it to use other CMFs when calibrating.
CIE 2012, 2 degree, needs an different WP then 1931 to get D65 for this color matching function.
The D65 xy cordinates for CIE 2012 is
x: 0.3135
y: 0.3308
Does this mean I can’t just punch in 6504 kelvin when using CIE 2012? Do I have to put in the coordinates instead?
Doesn’t DisplayCal/ArgyllCMS handle this internally depending of the observer? If I set 6504 kelvin when using the CIE2012 observer it should equal x=0.3135;y:=0.3308 or the temperature setting is hard-coded with CIE1931 and I have to set the coordinates manually?
Calibrite Display SL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Colormunki Display doesn’t have filters for other CMF’s. You need a colorimeter that is compliant for CIE2012.
You need a spectro if you want to calibrate in other CMF’s.
As far as I understand the ColorMunki Display can load colorimeter correction files for the specific display (CCSS,CCMX) which allows it to use other CMFs when calibrating.
Yes, you are correct, as long as:
-CCSS: filter spectal behavior and filter spectral data match (under some tolerable error).
-CCMX: you measure both reference device and i1d3 in CIE XYZ 2012 2 degree, otherwise computed matrix would be useless.CIE 2012, 2 degree, needs an different WP then 1931 to get D65 for this color matching function.
The D65 xy cordinates for CIE 2012 is
x: 0.3135
y: 0.3308
Does this mean I can’t just punch in 6504 kelvin when using CIE 2012? Do I have to put in the coordinates instead?
Doesn’t DisplayCal/ArgyllCMS handle this internally depending of the observer? If I set 6504 kelvin when using the CIE2012 observer it should equal x=0.3135;y:=0.3308 or the temperature setting is hard-coded with CIE1931 and I have to set the coordinates manually?
IDNK, set 6504 K daylight + 2012 2 degree in DIsplayCAL, then switch the WP combo to “xy” instead of color temp. If DisplayCAL is working as you want it will show xy coords of CIE 2012 2 degree D65 on “x” “y” textboxes.
IDNK, set 6504 K daylight + 2012 2 degree in DIsplayCAL, then switch the WP combo to “xy” instead of color temp. If DisplayCAL is working as you want it will show xy coords of CIE 2012 2 degree D65 on “x” “y” textboxes.
If I set the temperature to 6504 kelvin while using the CIE 2012 observer and change to coordinates it says is equal to x=0.3127;y=0.3291 aka CIE 1931 D65 coordinates:
If I do the opposite and set the CIE 2012 D65 coordinates and change it to temperature it gives me 6448 kelvin:
So I think you have to set the coordinates manually when using an observer that is not CIE 1931 but I am not sure.
My profiles using CIE 2012 6500K show in measurement reports around 6800K since the reports are hardcoded to CIE 1931, which seems to show it works. Sometimes switching between temperature and coordinates is finicky so yeah I would enter it manually and leave it just to be sure.
-CCMX: you measure both reference device and i1d3 in CIE XYZ 2012 2 degree, otherwise computed matrix would be useless.
There is no observer selection for the colorimeter. Only for the spectro. When generating a correction.
So just to reiterate, if I want a D65 whitepoint while using the CIE 2012 2 observer I should set the coordinates to x=0.3135;y=0.3308, correct?
-CCMX: you measure both reference device and i1d3 in CIE XYZ 2012 2 degree, otherwise computed matrix would be useless.
There is no observer selection for the colorimeter. Only for the spectro. When generating a correction.
Is it not using the observer selection in main screen? ArgyllCMS command has observer selection in ccxxmake.
$ ccxxmake Create CCMX or CCSS, Version 3.0.0 Author: Graeme W. Gill, licensed under the AGPL Version 3 Diagnostic: Output filname expected usage: ccxxmake -t dtech [-options] output.ccmx -v Verbose mode -S Create CCSS rather than CCMX -f ref.ti3[,targ.ti3] Create from one or two .ti3 files rather than measure. -d n Choose the display from the following list (default 1) 1 name = '\\.\DISPLAY1' 1 = 'DISPLAY1, at 0, 0, width 2048, height 1152 (Primary Display)' -dweb[:port] Display via a web server at port (default 8080) -dcc[:n] Display via n'th ChromeCast (default 1, ? for list) -d madvr Display via MadVR Video Renderer -d dummy Dummy (non-existant, invisible) display -p Use telephoto mode (ie. for a projector, if available) -a Use ambient measurement mode (ie. for a projector, if available) -y l|c Other: l = LCD, c = CRT -z disptype Different display type for spectrometer (see -y) -P ho,vo,ss[,vs] Position test window and scale it ho,vi: 0.0 = left/top, 0.5 = center, 1.0 = right/bottom etc. ss: 0.5 = half, 1.0 = normal, 2.0 = double etc. -F Fill whole screen with black background -N Disable initial calibration of instrument if possible -H Use high resolution spectrum mode (if available) -C "command" Invoke shell "command" each time a color is set -M "command" Invoke shell "command" each time a color is measured -o observ Choose CIE Observer for CCMX spectrometer data: 1931_2 (def), 1964_10, 2012_2, 2012_10, S&B 1955_2, shaw, J&V 1978_2 or file.cmf -s steps Override default patch sequence combination steps (default 1) -W n|h|x Override serial port flow control: n = none, h = HW, x = Xon/Xoff -D [level] Print debug diagnostics to stderr -E desciption Override the default overall description -I displayname Set display make and model description (optional) -t dtech Set display technology type (Use -?? to list technology choices) -U c Set UI selection character(s) -Y r|n Set or override refresh/non-refresh display type -Y R:rate Override measured refresh rate with rate Hz -Y A Use non-adaptive integration time mode (if available). correction.ccmx | calibration.ccss File to save result to
So just to reiterate, if I want a D65 whitepoint while using the CIE 2012 2 observer I should set the coordinates to x=0.3135;y=0.3308, correct?
Yes, DisplayCAL built in conversion seems to do not use selected observer.
Is it not using the observer selection in main screen? ArgyllCMS command has observer selection in ccxxmake.
In the main tabs yes. But not in correction generation screen for the colorimeter. Only for spectro.
It’s locked to the observer used for creation. So locked to 2012 in main screen when colorimeter correction is selected.
Yes, DisplayCAL built in conversion seems to do not use selected observer.
Even for lut generation also?
Yes, DisplayCAL built in conversion seems to do not use selected observer.
Even for lut generation also?
For profile generation? IDNK, but it should be easy to test: create a fast matrix profile and see which WP is stored in ICC. DisplayCAL stores actual white as WP not PCS.
For LUT3D generation I belive that it does not need an observer at all as long as all colorspaces used use the same coordinate system (refered to the same observer). I do not think even using argyllcms command line you can mix them (but I has not tested it)… so you may need to cook an alt Rec709 with CIE 2012 2 degree coordinate for an CIE 2012 2 degree display profile.
It is far easier to calibrate WP to whatever wp you want (including D65 2012 2 degree), then profile in CIE 1931 2degree. Then use in LUT3D rel colorimetric or if you want to cook your own rec709 profile then you can use abs col.I uploaded a picture. CIE 2012, 2° and CIE 170, 2° are the exact same thing when I verified them with a Jeti 1501 probe using two different softwares.
This picture is from Jeti’s cmf paper.
Interesting. Are these values mathematically calculated between the observers or empirically derived by calibrating a reference display with a reference spectrometer to CIE1931 D65 first and then measuring that white back with CIE 20112?
In either case, I would be interested in seeing CIE 2012 x,y coordinates for Rec709 and Rec2020 primaries (although, I guess the latter one is a bit too much to ask for in case these white x,y values are empirically derived). That way I could try using CIE 2012 overall for WOLED calibration instead of just using CIE2012 to derive a WP offset for CIE 1931 calibration.
Forgive me if this is a stupid question. I am honestly not sure if it’s possible to mathematically calculate x,y numbers between different observers. On one hand, my hunch is that it should be possible. On the other, I have never seen this done anywhere but I read about talented/important people doing the juggling between CIE 1931 vs. Jud-Voss 1978 using a CRT monitor to derive a white offset for OLED panels (for example, FSI recommends the “Jud-Voss method” or “Judd-Voss offset” for their RGB OLED displays -> Why would these people go through that trouble if they could just fire up Matlab/MathCAD/etc and get the numbers out of a copy-paste equation…? I also couldn’t find such equations with Google…).
P.s.: Now that I thought about this, a pure mathematic calculation might not be possible because technically an infinite number of spectral shapes could result in the same x,y values using either observer. If that is the case, it would be nice if somebody with a high-res spectro could calibrate a ~Rec709 gamut CRT/PDP display to quasy-perfect Rec709 primaries with a 3DLUT and measure back the x,y numbers using CIE 2012. The problem is, we will never find any CRT/PDP with Rec2020 primaries. Maybe we could try to extrapolate using the Rec709 white + the primaries…? I am a bit sad that Rec2020 was defined with CIE 1931 coordinates. The “observer metamerism” was a known problem when they finalized the new standard (and it was designed to be “future proof”, so it will probably stay with us for a very long time…). If they may consider amending Rec2020 with alternative observer coordinates, it would be wise to do so sooner rather than later.
Ah, I found the PDF (it’s public): https://www.jeti.com/files/content/support/downloads/papers/Lux%202017.pdf
Unfortunately, they didn’t make it clear how they derived this data (there is no mention of equations or what kind of display they used as a reference). I just remembered that D65 is defined spectrally (it consists of a table with the original points specified at 10nm intervals between 300 and 830 nm with linear interpolation used to get 5nm data as the standard D65 illuminant curve). Hence D65 coordinates can be calculated for any observer curves (that cover this wavelength range). However, Rec709 and other similar standards use CIE 1931 x,y values rather than a reference spectral curve, so those can’t be simply calculated.
It’s also unclear to me what they meant by “which white setting gave them the best neutral visual expression”. I mean… compared to what…? Did they also put a reference D65 white bulb there (one that is close to the spectra of the standard D65 illuminant)? They certainly don’t mention that if they did. If they just had to guess “which one of these do you think is white?” then I think this experiment was as good as asking them to randomly point at a display with fully covered eyes. they wrote “displays were switched quickly to avoid adaptation” but my experience says that’s impossible. My eyes/brain always pick one display (or whatever lightsource) to adapt to. That can’t be avoided by cycling all displays through settings. There is no “neutral setting” in our body. That’s not how it works. (And I say that despite how I can willingly choose how I see the internet-meme blue/yellow dress if I concentrate enough. I can swap between those two “settings of my brain” within a minute or so, just by concentrating. Nobody I ever talked about this believes me I can do that.)
By the way, this paper also concludes that no CMF can be recommended for general use on all display types.
Going back to the question of calculating vs. back-measuring white x,y for different display types… FSI says they measured the CIE 1931 D65-calibrated white-point of a CRT to yield 0.317, 0.341 with Judd-Voss. Now, this paper gives 0.3160 , 0.3351. These are close but not identical. So the authors of the Jety paper either used a different reference display (possibly also a different Jity model) for back-measuring (or even the Jety spectros have this much uncertainty in the 3. decimal point) or they used the tabulated SPD of the standard D65 illuminant to create that table. (Ah, well, I guess I should fire up a math software to tell.)
