5. The Circularity of Hue
4. Additive Colour Mixing
3. How We Describe Colour
2. How We See Colour
1. Colour and Wavelength
Short video about colour and wavelength
Why yellow and blue don’t make green
[and why we should stop teaching it in schools]
You will find images like the one above, that show that red, yellow and blue are the primaries and that yellow and blue make green.
Sometimes this is represented as a colour wheel:
So some people say yellow and blue make green. And you will find other answers that say that yellow and blue make black. How can this be?
Well, we need to understand a little science to get to the bottom of this.
The figure below shows what happens when you mix an ideal yellow dye with an ideal blue dye. The blue dye reflects light perfectly in about a third of the spectrum (and absorbs perfectly in the other two thirds). The yellow pigment reflects light perfectly in about two thirds of the spectrum (and absorbs perfectly in the other third).
The problem here is that the blue and yellow pigments (between them) absorb perfectly across the whole spectrum. The people who say that yellow and blue make black are saying so because of this argument.
Note that blue is a particularly bad choice of primary because it absorbs so broadly across the spectrum. [Making the blue even purer would only make the problem worse by the way.] Yellow is a good choice of subtractive primary because it only absorbs in one third of the spectrum.
The problem is, the people who say that blue and yellow make black are wrong of course. Every child knows this. In practice, if we measure the reflectance spectra for blue and yellow pigments they don’t look like those ideal ones I showed above. For a start, they are quite smooth. Here is a reflectance spectrum for a real yellow pigment. (The reflectance factor, by the way, is the proportion – or per cent – of light that the colorant reflects at each wavelength.)
Notice that with a real yellow colorant, it does not reflect perfectly in the middle and long wavelengths and it does not absorb perfectly in the short wavelengths. It reflects and absorbs to some extent all the wavelengths but it absorbs more at the shorter wavelength and absorbs at less the middle and longer wavelengths. The same is true of a real blue colorant; it does not absorb perfectly at the middle and longer wavelengths. The consequence of this is that you don’t get black if you mix blue and yellow. You would get black if the pigments were ideal but they are not. We live in the real world. However, you certainly don’t get a lovely bright green as shown in the colour wheel with red, yellow and blue primaries. You would get a dark desaturated murky dirty greenish colour. The main reason for this is that the blue is absorbing too broadly. Interestingly, if you look at the artist John Lovett’s page he explains that to mix a yellow and blue you should use a yellowish blue (and a bluish yellow).
Now let’s see what happens when we mix cyan and yellow dyes. We’ll start with the ideal colours.
It’s very nice. We get a lovely green colour. Cyan is a great subtractive primary because unlike blue it absorbs in only one third of the spectrum (the red or long wavelengths). Note that it is precisely because the cyan does not look pure that makes it a great primary – that’s why I get so furious about people saying the primaries are pure colours. The cyan looks bluish-green because it reflects in two thirds of the spectrum and only absorbs in the reddish part. Neither the cyan nor the yellow dye absorb in the middle (green) part of the spectrum and therefore the result of mixing cyan and yellow is a lovely green. Except it is not quite true. Remember, this is for ideal pigments. Real dyes do not look like that. Refer back to the measured reflectance spectrum for the real yellow pigment. In reality cyan and yellow do make green but the green might be a little less saturated than you may wish for because of the unwanted absorptions by the two dyes in the areas of the spectrum where ideally they would not absorb. (It was the great Robert Hunt, who worked for many years at Kodak – for those who knew him – who taught me about unwanted absorptions.)
Have you ever seen this happen. Of course, you have. Whenever you use a printer (which typically uses cyan, magenta and yellow primaries) to get a green, the printer is using cyan and yellow to make the green.
Remember those people who say that you can’t make blue because – yawn – it’s a pure colour that can’t be made by mixture? Well, have you ever printed out blue on a printer? Of course, you have. Let’s look again at our ideal primaries and see if we can explain it.
That’s right. Mixing cyan and magenta makes blue. The cyan absorbs in one third (the red third) and the magenta absorbs in one third (the green third) but neither absorb the short wavelengths.
John Lovett explains that you can do a decent job of mixing red, yellow and blue dyes, but only if you allow yourself to use multiple blues and multiple yellows, for example. If you want to do the best job possible using only three subtractive primaries, then the best you can do is to use cyan, magenta and yellow.
So finally you can see that the best subtractive primaries are cyan, magenta and yellow because the cyan is red absorbing, the magenta is green absorbing and the yellow is blue absorbing. And what is more, you now understand why this is the case (rather than accepting dogma). You also understand why there is a relationship between the CMY of subtractive mixing and the RGB of additive mixing.
The optimal additive primaries are red, green and blue (I will cover this elsewhere). And for this reason the optimal subtractive primaries are cyan (red absorbing), magenta (green absorbing) and yellow (blue absorbing).
But don’t be fooled by this lovely subtractive colour mixing diagram. You might not get such lovely blue, green and red colours when you mix real CMY primaries (either on your printer or with inks/paints). Why not? Because of the unwanted absorptions.
If you want to to know more you could do worse that get a copy of Measuring Colour, now in it’s 4th edition, and authored by Hunt and Pointer.
This post gets quite a few hits so I will take this opportunity to direct you to my short series of youtube clips that describe the issues discussed in this post in a visual way. You can see them here. If you want something a bit more technical check out this short lecture on colour primaries or visit my patreon.
Why the ‘three colour primaries’ rule is wrong
A great many textbooks state that there are three colour primaries. This is normally followed by the statements:
- All colours can be made by mixing together three primaries.
- The primaries – which are often cited as being red, yellow and blue – are pure and cannot be created from mixture.
Not only do I profoundly disagree with these last two statements but I disagree with the statement that there are three colour primaries. Here’s why:
It is relatively easy to go into a lab or studio, start with three colours (any three; you pick’em) and find that you cannot make all colours. People who do this will often say, that theoretically you can make all colours from, say, red, yellow and blue but that practically you can’t simply because the primaries are not pure enough. The problem is, the more pure you make the primaries, the fewer colours you can make!! The fact is you cannot make all colours from three primaries no matter how carefully you choose the primaries. You cannot do it practically and you cannot do it theoretically.
We can trace the idea that primaries are ‘pure’ back to ancient Greece. In those times and for centuries afterwards it was even frowned upon to mix colours at all because of the loss of purity.
It turns out that if you want to make a large range of colours using three inks or paints, the primaries you should choose are cyan, magenta and yellow. Don’t just take my word for it. Go and ask HP, Canon or Xerox. These companies have made printers for decades and make a living out of selling devices that allow consumers to make a wide range of colours with just three primary inks. They all use cyan, magenta and yellow as their primaries.
But how can magenta be a primary you might ask? It’s far from pure. That is because the notion of primaries being pure is an outdated idea (outdated for several centuries I might add) and should not be taught in Schools. Cyan, magenta and yellow make good primaries for an ink system precisely because they are not visually pure – they each absorb in a narrow part of the visible spectrum and therefore emit light quite broadly. Blue would make a poor primary in an ink or paint system because it absorbs at too many wavelengths. Mixing together blue and red inks make a very dirty brownish black colour. So the gamut (the technical term for a range of colours produced by some primaries) of colours we can make from red, yellow and blue inks or paints is quite small.
So, we can’t make all colours from three primaries, the best primaries are not those that are pure, and primaries can be made by mixing other colours. It is easy to show that a blue can be made by mixing together cyan and magenta inks and this is shown rather nicely by the artist Scott Naismith in this very nice youtube video.
We tend to use three primaries in many systems because you can make a great many more colours with two primaries than with one and you can make a great many more colours with three primaries than you can with two. But you can’t make all colours with three. You can make more colours (a larger gamut) with four or five primaries – though you still can’t make them all – but we reach the point of diminishing returns. Is it worth the extra expense of having four or five primaries if three do a pretty good job? Usually not. However, sometimes we do think it is worth using more than three primaries. For a start, most printers use CMYK (cyan, magenta, yellow and black) so that is four primaries. Then we have hexachrome printing systems with six primaries. The Quattron TV is manufactured by Sharp and has four primaries (red, green, blue and yellow) whereas most TVs only have three (red, green and blue).
The truth is there is no perfect set of primaries and there is no fixed number. A set of primaries is simply a set of colours in a colour system that can make a useful range of colours (gamut). Very often three hits the commercial soft spot but that’s just about engineering and economics.
For further information list to my podcast about colour
Or visit my patreon page at https://www.patreon.com/colourchat
What type of colour information do designers want?
In this study we were interested in which type of colour information designers want. We carried out surveys and interviews (with senior designers and brand managers) and the results are summarised below:
We used a card-sorting technique in our interviews to ensure that the participants knew what each of our terms meant.
We found that colour meaning was one of the aspects of colour that designers would like to be able to put their finger on; it was more important that colour trend information in fact! We also looked some existing colour tools and found that none of them really offered the most important information that designers and brand managers want to know about colour. What would be really cool would be a tool that provided accurate information about the meanings that colours have in different cultures and perhaps in different contexts.
The full paper will shortly be published in Color Research and Application.
Won S & Westland S, 2018. Requirements capture for colour information for design professionals, Color Research and Application.
The full publication details will be added here when they are available. Meanwhile, you can read it here.
would you like pink chocolate?
About 80 years ago the Milkybar was introduced by Nestlé. Since then, chocolate has broadly speaking been one of three colours: dark, milk and white. Today I read that a new colour of chocolate has been developed which is claimed to be the first new natural colour of chocolate since Nestlé’s innovation. The beans are grown in Ivory Coast, Ecuador and Brazil and the new chocolate, which is being referred to as ruby chocolate, has been underdevelopment for just under a decade. Apparently this new chocolate has a natural pinkish colour and a berry flavour. I suspect the manufacturers are choosing to call it ruby chocolate rather than pink chocolate because the latter sounds childish; they probably want to market this new chocolate in the upper price brackets and emphasize that its colour is natural (there are plenty of pink children’s sweets out there that are full of artificial colorants).
