5

Munsell color reference chips are really useful for painters but awfully expensive. It should be easy to create a sheet that can reliably be printed on most printers though, shouldn't it? I.e a PDF with vector images (to avoid rasterization artifacts) of colored squares (with Munsell labels) that comes from a regular sampling of some volume of CIE Lab (we don't really need the really saturated chips, matching near neutral colors is more useful).

One problem is that most color profile workflows deal with matching the monitor to the printer, while what we want to do here is send the right CIELAB values so that two sheets from the same printer match (assuming their CIELAB->CMYK profiles are correct). Vector formats like SVG do not yet reliably implement specifying colors in colorspaces like CIELAB.

I suppose one can just start by sampling CIELAB and then converting to a standard RGB colorspace then printing. I am not sure if this could be problematic with two hops.

Is there a more clever way to create such a color reference? One would think it would have been done long ago if it was this easy. Is there something about reproducibility between printers I am not taking into account?

EDIT: Paul Centore has some wonderful tools and write ups on his site:

http://www.munsellcolourscienceforpainters.com/

He get right to the main question in this very accessible and detailed writeup (with full code posted on his site): http://www.munsellcolourscienceforpainters.com/HowToPrintAMunsellBook.pdf

Why doesn't this quite answer the question? His writeup assumes that the printer is a blackbox and he calibrates output with the X-rite "ColrMunki" spectrometer. If we can trust the printer ICC profiles that map CIELAB->CMYK then we should need to do this if we (implicitly) send CIELAB values to the printer. I would have assumed (for high quality printers) the ICC profile is reliable, is that the case?

UPDATE: A personal communication with Paul suggests that indeed, color profiles typically do not predict color coordinate values within 1DE (a measure of colorimetric distance). Colors that are farther apart than 1DE are generally distinguishable by the human eye. Some profiles and commercial printing processes are often 4-5DE away and can be considerably worse in the worst case. Paul's approach, using a spectrophotometer, was demonstrated to often predict under 1DE (using an Epson R2880).

It appears that despite considerable rigor in printcraft, we do seem to be limited inherent variability in commercial printing processes. While color profiles should enable us to reliably print Munsell color charts in principle, the underlying physical variance in the process seems to swamp out the predicted gains in accuracy that might be obtained with better sampling and interpolation from profiles.

  • 3
    I think you've answered your own question with "it would have been done long ago if it was this easy". It would be very difficult, very time consuming and the results would most likely be unsatisfactory. – Westside Jan 6 '17 at 8:41
  • 2
    Many of the colors in a real thing simply cannot be printed by generally available low cost printers . The available variety of different colors is insufficient. The common term for this: Out of gamut. Computer screen can have larger color space, but it's mindless to compare screen colors and painted colors because the result of the comparison depends heavily on the brightness of the screen and how the painting is illuminated. Painting can have the same light as the Munsel sheet nearby – user287001 Jan 6 '17 at 9:59
  • My former comment was based on non-positive thinking. I try to correct. So, if a predefined small group of people want to talk or write about painted colors, it's well possible if they had identical color charts. That chart (definitely not Munsel) can be printed by using the same low cost printer if it's color space is large enough for the paintings under the talking . – user287001 Jan 6 '17 at 11:14
  • I'm puzzled as to what I am missing. In principle, the ICC profile maps image colors to printer CMYK amounts through a profile connection space like CIELAB or CIEXYZ. As long as those profiles can be expected to be accurate (I have no idea if they are for, say "good" office printer models); this is, in principle, possible (there are straightforward ways to transform between RGB, CIELAB/XYZ and Munsell). The "out of gamut" aren't really relevant, as I already said that I am more interested in neutral colors. – mittimithai Jan 6 '17 at 19:44
  • Actually you have missed nothing in theory. I still haven't got similar looking outputs of the same photo from different printers altough color management is at least tried to be in use. That leads me to assume the available color profiles to be quite coarse. Btw. I don't have a formula or table on how to transfer Munsell Hue, Saturation and Brightness indexes to RGB or CMYK values in some printing environment. Have you any idea how to get it? I've not called to Munsell Color Company and asked them to send one. – user287001 Jan 6 '17 at 21:15
0

There seems to be plenty of demand of low cost Munsell charts. This subject has been discussed a long time. There exists software for conversion between Munsell indexes and usual color models, low cost charts in sale and free chart PDFs.

Ihave no idea,

  • are they properly color managed or only for teaching the system?
  • which easily available printers are up to the task; even if high chroma color chips are not used

See some of the PDFs; link= irtel.uni-mannheim.de/colsys

Maybe you can print one and compare the result if a real thing is available even for a short time.

Late addition: There's published lists of equivalent computer colors of a set of Munsell chips. They are claimed to fit in a certain standard light. The following is obviously the result of an academic work to calculate the equivalents. They are not measured from Munsell Book of Color.

http://www.rit-mcsl.org/MunsellRenotation/real_CIELAB.xls

I do not know who can print them acceptably so that side by side they look the same as the real thing. They are computer colors, not pigment receipes. Only pigment receipes can have a theoretical possiblity to be the same as the real thing in every light.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.