Colour (calibration) matching from a 4-panel colour (checker) reference

I feel like your card which is perhaps printed on plastic?? which might yellow overtime?? Is really only good as an internal normalizing reference between photos and only likely for ones taken with the same card… I just quickly downloaded your image and maybe it was the export but if you balance on the white the grey was quite blue and if you balanced on the grey the white became much more yellow/warm… out of the gate the red green and blue were not badly lined up with those hues on the vectorscope but yellow was off and the saturation of green was considerably less than the other patches…not sure if that would be an issue but I can 't help but question the consistency of a 2 dollar card… So for me the best use case would be using it as an internal normalizing standard or reference that will not necessarily give you accurate colors but might give you a way to adjust a set of images taken with the same card to the same point…

Hey all, lots of great discussion here and useful answers, and I’m a newbie on the forum so I’m capped at how many people I can reply to directly, so kind of a general answer.

But essential, yes, @Claes is right - we use these cards in lab and field contexts for archaeological recording. I understand from all the answers that without a set calibrated setup (which is definitely not the case - we have several different models of field camera, and a couple of different lab cameras, and most of my colleagues use the cameras on Auto mode), so the best use of these cards for me will be as a reference tool - from what I understand, something I can check to make sure the images have been processed correctly.

For lab contexts, e.g. cataloguing and photographing large collections of artefacts, I have introduced my own setup which involves strictly controlled lighting, camera calibration, and a 24-panel colour checker, so there’s no real concern for lab imagery going forward. But as far as the existing archive of images, I understand I won’t be able to calibrate based on the four-panel cards.

@priort that’s not my actual card - just the photo from the webstore linked in the question. We have a large collection of cards printed to the same calibration spec on the same date that we can access, so that we don’t use damaged, faded, dirty, or miscoloured cards. But at the end of the day, since we work in very remote conditions and pack very light, we have (so far) been trading off the accuracy of top spec colour calibration equipment in favour of these.

Thanks everyone for so much information, it’s been very helpful! Og @Claes, kan du sende meg meldingen som du sendte til Terry, vær så snill? Jeg er interessert, og også en brukere på stackexchange har spurte de samme sporsmål og jeg ønsker å kunne svare (eller jeg kan gi jeg URL-en hvis du vil spurte!) :slight_smile:


Check your messages!

1 Like

Thanks for bringing up an interesting subject. :slight_smile:

Sounds like a plan.

Hi Claes,
I agree that sharing the very informative message that you sent me would add to this conversation thread. I thought that white balance alone would preserve correct colors and you provide evidence that this is not the case.

1 Like

@paperdigits perhaps a more relevant example that I’d forgotten to include is this rock art colour calibration card.


As you can see, it’s a non-standard calibration card, but has plenty of reference panels. Do you know how something like this might be used?

1 Like

You really aren’t going to get anywhere useful thinking about device dependent RGB and CMYK values. To actually profile a system you need to be dealing in device independent values, so that you have a basis for relating device dependent colors back to them.
i.e. first step is to measure the reflection spectrum of your reference card using a spectrometer. You then have some idea what these colors should be in some other defined colorspace such as sRGB or something else that you have a color profile for. Creating a useful conversion is a further exercise, but at least you would know what you are trying to convert from and to.


@gwgill So is it correct to assume from this that an Xrite/Calibrite ColorChecker has known values assessed by a spectrometer, and that in e.g. Darktable where they can be natively handled, the correct values are known for all available colourspaces?

I’m not a colour scientist or expert so this is possibly a bit of an ignorant question, but if a printed colour has a known CMYK value, is that not the same as knowing the value from a spectrometer? I have access to multiple spectrophotometers (is that not what you meant, rather than a spectrometer?) so I can check the colour values - but the outputs from a spectrophotometer are RGB, CMYK, hex, etc. which I already know for this card. How is that different? :confused:

1 Like

Very, very far from the same thing. CMYK values are the notional amount of ink laid down. What color that actually makes depends on the nature of the inks, the way they are screened, the paper they are printed on, and the light being used to illuminate the print. All of that is highly variable with every component involved. Similarly with RGB values. They are device dependent - i.e. the color they produce depends on the nature of the device used to display them.

The raw output of a spectrometer is the percentage reflectance at each wavelength. Add an illuminant (say D50 by default) and an observer model (say the 1931 2 degree standard observer) and you can compute the device independent XYZ values that represent the color. That’s the common “language” to connect device dependent colors.


My findings are presented here:

Have fun!
Claes in Lund, Sweden