Understanding Color Spaces and the CIE Chromaticity Diagram

Ya’akov

Joined Jan 27, 2019
6,856
Recently, in doing some research on the 1931 CIE Chromaticity Diagram¹ I ran across this document from Gernot Hoffman². It is an information packed paper on the CIE color space and so also touched on the nature of the human gamut, other color spaces (sRGB, Adobe, etc.) and the systems used to describe color mathematically.

1931 CIE Chromaticity Diagram
(source)​

The paper is very information dense, and it is written in German-influenced English. The combination can make reading it a bit of a slog, but it is very informative about both theory and practice concerning color space. It even has Pascal source code for some programs (I found this quaint, but it’s more than I did!).

It was very helpful in consolidating my understanding of some aspects of color spaces, something of interest to me because of photography—but it also has special value in understanding LED specifications and binning. There is also no small value in realizing how dubious the idea of “color” is. Much of what is used to specify color doesn‘t “exist” because only various approximations are possible.

It also drives home how different the human gamut is from any used for reproducing color, be it by additive illumination or subtractive printing. If you have an interest in color, or haven’t really thought about it, give the paper a scan. There is a fair bit of math there too, not my strong suit but very useful in practical terms.

1) It’s called “1931” because that was the year the system was established by the Commission Internationale de L'éclairage (International Commission on Illumination), the French being responsible for the CIE initialism, like UTC for ”Coordinated Universal Time” coming from the original French name. Amazingly the experiments that lead to the system were even older, being performed in the 1920s!

2) source: http://docs-hoffmann.de/ciexyz29082000.pdf

Attachments

• 1.6 MB Views: 2

DickCappels

Joined Aug 21, 2008
9,512
The 1976 CIE L* u* v* color space is more uniform and better for describing hue differences. It works the same way as the 1931 space but deltas in that color space seem to give the impression regardless of where they are in the color space. The formulas are, of course, different.

https://en.wikipedia.org/wiki/CIELUV

Ya’akov

Joined Jan 27, 2019
6,856
The 1976 CIE L* u* v* color space is more uniform and better for describing hue differences.
Yes, thanks for pointing that out. But there are still a lot of things that reference the 1931 space for whatever reason they have.

DickCappels

Joined Aug 21, 2008
9,512
It is about the same reason every new electronics student tries to use an LM741 in their experiments. If memory serves the LM741 was the first internally compensated opamp and appeared in the text books and peripheral literature early on, and it stuck.

1931 color space is fine for most things, and for the most part people are very comfortable with 1931 color space.The 1976 color space was adopted because the need for a more uniform color space arose.