CIE Chromaticity Diagram

|

Chromaticity Diagram

The chromaticity diagram that we use for talking about color spaces and color calibration looks simple. And that’s the power of it really. Its simplicity is genius. Let’s dive into it a little bit with a few basic understandings. My intent in this post is to cover the basics and make it understandable for photographers with how it applies to them. I fully understand there’s way more science behind the scenes that I’m pretty sure I can’t begin to comprehend, nor do I have the interested in comprehending it. However, I value a basic understanding of these types of items because it does enhance our ability to successfully execute color management and to understand where standard terms are coming from so we are consistent. With that said, let’s get to it.

The shape of it

The shape of the diagram is somewhat curious to me. I saw one reference to it saying it looks like a “tongue” which I thought a bit odd. I prefer to say the “horseshoe shaped item” when referring to the diagram. Even that’s not perfect, maybe someone was on to something when they called it a tongue.

What we see is two axis, the Y axis is the vertical and the X axis is the horizontal. In this model we don’t see the Z axis which refers to luminance values. Which brings an interesting reality to understanding this diagram. We humans perceive greens to be much brighter than the violet and red counterparts. And that seems to be the reason why this diagram is stretched and so much green is represented. However, this diagram effectively corrects for that perception and equalizes the intensities or luminance values inherent in these colors. This diagram was created in 1931 and there have been modifications to it to better reflect human perception but this remains the standard for all color calibration systems that I’ve used.

The outer line represents colors at their most pure. I probably should put it more scientifically, that is it represents light as pure wavelengths measured in nanometers (nm). So the representation of the diagram tries to equalize the luminance of the colors, in that they are treated equal in this plane, but let’s remember perceived luminance is the Y axis. The best description I could find for the X axis (the horizontal axis) is that it’s a “mix of response curves chosen to be nonnegative.” This is the part where I accept what they’re telling me and I don’t try to push my brain too far. My interests in in understanding just enough to get accurate and repeatable color out of my screens and printers. However, I should also take the opportunity to thank the people that do understand all these scientific applications and give us the tools in easy to use packages so we can have great and consistent color.

One last thing on the shape of the diagram, notice the straight line along the bottom portion. It’s known as the “line of purples.” Anything along this line can be defined as a purple so long as it’s not pure violet, about 420nm, or pure red, about 680nm. Any color along the line is achieved by various amounts of the two opposing light values, a little red and a lot of violet will deliver a result closer to the violet side of the diagram and vice versa. 

Color models vs color space

Something I learned in my research for this article was the proper use of the term color space. We always use this term to reference something like ProPhoto RGB, Adobe RGB, sRGB and so on. In reality, these are color models. Though the term color space has been accepted in the industry, this hasn’t always been the case. I still plan to use the term color space, but I wanted to at least address this issue. For the purposes of this article, I’ll continue using the term color model to refer to what we commonly call a color space.

The diagram represents the true color space. This is the color space that represents the average of human vision as it relates to color sensitivity. All other color models are a subset of this color space. What I find very interesting about the relationship between the color space as represented by the diagram and any given color model is the limitations of those color models. It’s also great to understand how they work together. That is when you take a look at ProPhoto RGB and the green starting point, it’s actually an “imaginary” color, as is the blue or violet starting point. That is, it’s outside the spectrum of what we can see with our human visual system. However, because it’s so far out, as a starting point, it is able to encapsulate so many more colors that we can see. The human visual system cannot be defined by a triangle, it’s more fluid than that. But our digital systems have to be represented by a triangle shape since we have three starting points defined, one of the green, one in the blue and one in the red. If you look at my referenced link you’ll be able to see the plotted coordinates for the various RGB starting points on the CIE diagram for the ProPhoto RGB color model. 

Color values vs perception

It’s important to note that this diagram has its limitations. Its purpose is to establish a framework in visual form, but as it’s only two dimensional and it doesn’t necessarily address perception, it has limitations. But this is also a strength when it comes to establishing calibration for our screens and printers. It does include all values of colors, those emitted and those reflected. It does not, however, take into effect variations between different people and how they perceive color. The researchers tested many subjects when doing the studies that led to the development of the diagram. Additionally, the diagram only shows purity of the colors at their specific measurement in nanometers and the cross-section of what happens when you blend the three primaries of Red, Green and Blue. That’s why the central part is desaturated. 

And finally, perception of items around us depends on the light shining on it, so if you were to describe something as being a certain color and identifying that color on the diagram, it would be different if you lit that same subject with a different light source. Yet the human visual system would likely compensate for those color differences and this phenomenon is called “color constancy.” Our digital cameras have “auto white balance” which is somewhat similar to color constancy, but alas, it is simply an adjustment of what we call a “white point” and is determined by the predominance of colors usually measured in degrees Kelvin. I’m diving off into a whole other subject, so let’s get back to the main topic here.

Additive vs Subtractive Color

Earlier I’d stated that the diagram encompasses all color models. This would be any of the RGB and CMY based models. We usually call the latter a CMYK model since blacK is needed to give the final printed result a rich appearance. In today’s inkjet printers we have way more than the CMYK inks. Often we’ll find a Light Black, Light Light Black, Orange, Red, Green and other inks as well such as Photo Cyan and Photo Magenta. These are all variations of a base CMYK color model idea. That is, they rely on reflected light in order to convey their color.

So why the term “Additive” and “Subtractive” for our color models? The reason is actually quite simple. You come about the answer to that by answering one other simple question. That is, by what method do you create the presence of color? In an RGB color model, light is emitted from a source. You can have separate emission points such as in a computer monitor with differently colored pixels, or you can have something like a light bulb or the sun. Either way, light is emitted from the source. So therefore, light is added, and color is added before anything can be perceived. 

With the CMY color model light is communicated through an act of absorption. So light or color is subtracted from the reflection of the original light source. That means when you look at something that is red in color, it’s because all other wavelengths of light are being absorbed by the pigments or colorants found on the surface of the item.

There are limitations to the CMY primaries and that’s why we have so many additional inks in today’s inkjet printers. The manufacturers are simply trying to expand the possible color model for their printer so you’ll want to buy it. For example, my Epson printer has Orange and Green inks. You’d think this would make the prints way better than my Canon printer that doesn’t have those inks. In reality, it’s not really noticeable. Plus, those Orange and Green inks are virtually never used. I’ll change out all other inks way before I change out those two on that Epson printer. So I suppose it’s nice to have, but largely unnecessary.

My Canon has a Red ink which I though interesting since red is one of the easiest colors to convert from RGB to CMYK.

Conclusion

The CIE 1931 color space chromaticity diagram is a very useful tool in understanding the relationship of color models and the human visual system. There’s so much science and research behind it that I won’t ever pretend to understand it all. However, a basic level of understanding is required for proper color calibration which provides consistency across devices. 

References

Diagram with nm plotted: https://en.wikipedia.org/wiki/CIE_1931_color_space#CIE_xy_chromaticity_diagram_and_the_CIE_xyY_color_space

ProPhoto RGB reference: https://en.wikipedia.org/wiki/ProPhoto_RGB_color_space

How Color Works available wherever books are sold. https://www.amazon.com/How-Color-Works-Twenty-First-Century/dp/0190297220/ref=sr_1_1?keywords=how+color+works&qid=1560099478&s=gateway&sr=8-1