Colour

Colour is slippery

It is hard to talk about colour. It turns out the words I usually use don't quite mean what I thought they did. Saturation is not a synonym for chroma or colour purity nor is brightness a synonym for lightness, luminance, value or tone (despite the assertions of my thesaurus).

And if at any time I speak of Light and Rays as coloured or endued with Colours, I would be understood to speak not philosophically and properly, but grossly, and accordingly to such Conceptions as vulgar People in seeing all these Experiments would be apt to frame. For the Rays to speak properly are not coloured. Isaac Newton - Opticks

For more (better) colour information I recommend The Dimensions of Colour by David Briggs (and particularly his glossary).

Light

I apologise that I'm going to talk about mixing light before I talk about mixing paint. They are related but different. Don't get them confused.

Light doesn't so much have a colour as much as it has a wavelength. Or rather individual photons have a wavelength. But light is made up of many photons of possibly many different wavelengths.

Light with a mixture of wavelengths has a spectral power distribution. Photons can be separated by wavelength using a prism.

Colour is what we perceive when that light excites our eyes. It is our minds interpretation of those wavelengths.

Each wavelength triggers a detection response and it is the perception of the combinations of detections that is the colour.

If the colour that we perceived was green then most people would also say that's green light, but some would disagree.

There are three types of colour receptor cells in the eye, the cone cells. Their sensitivity ranges overlap so light of a single wavelength may trigger more than one type of cone cell. The hue perceived is determined by the ratio of the types that are triggered.

But exactly the same response can be triggered by many different mixtures of wavelengths.

Sodium lamp

When light of wavelengths near 589nm come into our eye we might say:

See that sodium lamp? That's yellow.

But there are also mixtures of green (520–560nm) and red (700–635nm) lights that can cause exactly the same response.

No yellow pixels here.

See that particular mixture of red pixels and green pixels on the TV? That's yellow too.

There are no yellow pixels in a TV screen.

This phenomenon of multiple different possible causes triggering the same response is called metamerism.

Metamers - the same perceived colour from different spectrum of wave lengths that led to it. In both cases the ratios of short, medium and long wavelength sensitivity cones being stimulated would be the same. Our eyes can only report the rates that each type of cell is firing and their relative rates give us the hue. Our eyes can't tell us exactly what wavelengths caused that stimulation ratio.

Any yellow we see on a CRT monitor or TV is always a mix of green and red. They don't have any yellow emitters, only red, green and blue. To our eyes any two metamers are completely indistinguishable.

Magenta

More bizarrely we can say:

See that mixture of red and blue? That's magenta.

The perception of magenta cannot be produced by light of any single wavelength. Magenta isn't anywhere on the spectrum of visible light. It is just our perception produced from certain ratios of wavelengths.

The hue we perceive is the ratio of receptor types firing. Just because no pure light of single spectral wavelength can produce this ratio doesn't make the colour any less real.

A colour that can be produced by light of a single wavelength is called a spectral colour. Those that can't are called extra-spectral. Fair enough. Extra-spectral colours are all caused by mixtures of spectral lights - but there is no magenta, or pink, or gray wavelength - we can't have individual magenta, or pink, or gray photons.

A colour produced by a single wavelength can be identified by that wavelength. Metamers for those colours can also be associated with that same dominant wavelength even when there is no light of that wavelength present in them.

If we call 589nm sodium yellow and then find that the mix of 58% 542nm plus 13% 715nm plus 29% 785nm stimulates the exactly same ratio of cones, and so looks exactly the same, we could equate that mix with sodium yellow and say it has a dominant wavelength of 589nm even though it doesn't contain any 589nm light at all.

The mix has the same colour but not the same spectral power distribution. (That particular example mix probably doesn't look the same as I made the ratios up).

Note the ratio of receptors triggered by some light is not the same as the ratio of receptors triggered by the average wavelength of the light. It's a little more complicated than that. The average of red and blue wavelengths would be green but we see magenta.

This is the sort of thing that leads some people to say:

Light doesn't actually have colour. Objects don't have colour either.

They are right but it's sometime just sophistry. Colour is the perception. If some light consists of a mixture of wave lengths such that anyone with normal vision would say I see that as yellow then it seems a bit perverse to refuse to call it yellow light.

If there's enough different wavelengths and they are sufficiently evenly spread…

See that mixture of assorted wavelengths? That's white.

It's important to point note that white is extra-spectral. There is no white wavelength. There are no white photons. It is always a mixture of other wavelengths.

It doesn't need all the wavelengths to be present. That mean's some white lights are different to others.

Any white displayed on a CRT screen is actually from red, green and blue emitters. Other options are available, it has many metamers.

Irrelevant now but: Later, when I say white light I mean a full spectrum white. Like it was emitted from a black body at 5500K.

Even more irrelevant: the black in black body radiation is talking about what the body can absorb. The fact that it's also emitting light makes the name really confusing. 😦

Bulbs have a Colour Rendering Index, a percentage rating which indicates how full their spectral power distribution is. For lighting a studio always go for bulbs with a high CRI. And a high colour temperature. 5000k plus.

Sunlight and filament bulbs are black body sources, in general, but not fluorescent lamps. Fluorescent lamps, even those with good CRIs, can have noticeable gaps in their emission spectrum.

And finally:

See that mixture of no light? No? Neither do I. That's black.

Black is a colour.

The dimensions of colour

The CIE defines a colour space as a geometric representation of colour in space, usually of 3 dimensions. CIE 17-22-059 colour space

HUE: "attribute of a visual perception according to which an area appears to be similar to one of the colours: red, yellow, green, and blue, or to a combination of adiacent pairs of these colours considered in a closed ring" (CIE e-ILV*, 17-22-067).

Colour attributes perceived as belonging to OBJECTS:

LIGHTNESS: "brightness of an area judged relative to the brightness of a similarly illuminated area that appears to be white or highly transmitting" (CIE e-ILV, 17-22-063).

CHROMA: "colourfulness of an area judged as a proportion of the brightness of a similarly illuminated area that appears grey, white or highly transmitting" (CIE e-ILV, 17-22-074).

Colour attributes relating to appearance of LIGHT:

BRIGHTNESS: "attribute of a visual perception according to which an area appears to emit, or reflect, more or less light. NOTE The use of this term is not restricted to primary light sources" (CIE e-ILV, 17-22-059).

COLOURFULNESS: "attribute of a visual perception according to which the perceived colour of an area appears to be more or less chromatic" (CIE e-ILV, 17-22-072).

CD2023-1: "Controlling Colour: Historical Background" with Dr David Briggs

Physical colour models

Physical models often use dimensions which suggest how the colour could be constructed.

The RGB model uses the dimensions of how much red, green and blue light would need to be combined to create the colour.

Many different colour spaces such as sRGB, Adobe RGB, DCI-P3, ACES, etc. are built from the RGB model by defining some specifics for the dimensions.

The CMY and CMYK models specify how much cyan, magenta and yellow, and for CMYK: black, ink, dye or paint would need to be combined to produce the colour.

Black is not strictly necessary as a mix of the other three usually produces a fairly decent black, but printers are fussy about getting really dark darks. Besides, the other inks/dyes are more expensive.

K is used for black because B stands for blue.

Again, many colour spaces have been created from these models by specifying slightly different definitions for the dimensions.

Perceptual colour models

Perceptual models use dimensions that relate more to the human vision. They try to model the way in which the colour is perceived rather than how it's created.

The clue is in the name. Most use hue, chroma, and lightness as their dimensions or some variants of those.

In the light section above, most of what was said was only about one aspect of colour: its hue.

Hue Represented by angles around the central value axis.

The closest colour on the colour-wheel.

Hue: The attribute of a visual perception according to which an area appears to be similar to one of the colours: red, yellow, green, and blue, or to a combination of adjacent pairs of these colours considered in a closed ring (CIE, 17-22-067 hue).

Hue is determined by the ratios of the firing rates of the different types of cone cells.

Hue scales can be based on the dominant wavelength of the closest spectral colour, red, orange, yellow, green, cyan, blue and violet, or some analogous quantifier for extra-spectral colours such as magenta (eg. the complementary wavelength, the dominant wavelength of its complementary colour).

Value, brightness, lightness or tone Represented by distance vertically parallel to the central axis.

A ranking of the amount of light present.

In some spaces the dimension relates to the absolute measure of excitation and in some the term refers to the relative perception of a particular area in comparison to a similarly illuminated white.

It is determined by the total combined excitation levels of all types receptor cell.

The HSV model uses the term value differently. Colours with high values on the scale also include fully saturated colours rather than simply those close to white. HSL lightness runs from black to white but HSV value is scaled to run from black to the maximum possible for that hue and chroma. I avoid using HSV because its confusing. 😦

Brightness is our perception of luminance (visible light energy),

Brightness: attribute of a visual perception according to which an area appears to emit, or reflect, more or less light. NOTE The use of this term is not restricted to primary light sources (CIE, 17-22-059).

Lightness indicates how dark or light a colour is.

Lightness: (of a related colour) brightness of an area judged relative to the brightness of a similarly illuminated area that appears to be white or highly transmitting NOTE Only related colours exhibit lightness. (CIE, 17-22-063 lightness).

Lightness is how light a colour appears in relation to others. It's not an absolute measure of how much light is coming off. It's nearly a measure of how much light is reflected except that it can apply to TV screens and monitors. Lightness is thus "perceived reflectance" [Arend, 1993]

Saturation, chroma, intensity or colourfulness

Saturation indicates colour strength, the purity or intensity of a colour.

Indicates how far it is from being a grey.

The Difference Between Chroma and Saturation
Dr David Briggs

Saturation colourfulness of an area judged in proportion to its brightness

NOTE For given viewing conditions and at luminance levels within the range of photopic vision, a colour stimulus of a given chromaticity exhibits approximately constant saturation for all luminance levels, except when the brightness is very high. (CIE, 17-22-073 saturation).

Chroma colourfulness of an area judged as a proportion of the brightness of a similarly illuminated area that appears white or highly transmitting

NOTE For given viewing conditions and at luminance levels within the range of photopic vision, a colour stimulus perceived as a related colour, of a given chromaticity and from a surface having a given luminance factor, exhibits approximately constant chroma for all levels of illuminance except when the brightness is very high. In the same circumstances, at a given level of illuminance, if the luminance factor is increased, the chroma usually increases. (CIE, 17-22-074 chroma).

These are measures of the spread of the rates that the different types of cone cells are firing.

If one or two types of cone cell are quiet and the others are firing that can be perceived as a saturated colour, whereas if they are all firing fairly evenly then that is perceived as a desaturated colour.

Based on their absolute rates that gives us black, when none are firing, through grey when there is a balance to white when they are all firing rapidly.

Slight imbalances give us subtly coloured greys. Beige, bone, eggshell, almond. Further imbalance gives us browns and pinks, olive drabs and khakis and Prussian and powder blues.

Spectral colours, red, orange, yellow, green, cyan, blue and violet, are of high chroma where as multi-wavelength colours such as pink or beige, grey or white are low chroma.

Chroma type dimensions can be scaled so that the maximum chroma extra-spectral colours such as magenta or purple have the same value as a genuine spectral colour.

In some models the chroma dimension may also be known as the saturation or intensity of colour. In each case the low end of the scale is a grey (which may include pure black or white) and the high end is as colourful as can be. For spectral colours the high end will be a single wavelength colour.

I may have mentioned before that I avoid using these spaces because they suck. 😦

Chromaticity

Chromaticity diagram: an xy slice though CIExyY

Chromaticity is an indication of hue and saturation without brightness. It provides a qualitative description of a colour without mention of its value.

Chromaticity diagrams can give us a better feel for how the spectral and extra-spectral hues form the closed ring from which the the artists' colour wheels have been formed.

Chromaticity: The property of a colour stimulus defined by its chromaticity coordinates, or by its dominant or complementary wavelength and purity taken together (CIE, 2011 17-23-052 chromaticity).

In the chromaticity diagram we see a horseshoe shaped area which represents the gamut of human vision. The curved locus at the top is where the pure single wavelength colours would lie. The straight line at the bottom is the line of purples made from fully saturated extra-spectral colours. The triangle in the middle shows the colours that can be encoded in the RGB space and represented on a computer screen.

The colours used for the gamut edges are just approximations because they lie outside of what can be show on a computer!

Perceptually uniform colour spaces

A perceptually uniform colour space attempts to make the difference between two colours (as perceived by the human eye) proportional to the Euclidian distance between the points within the space. CIELAB does a fair job in any local area but not so well as the distances increase.

Over the years the CIE has defined improved colour difference distance measures (Delta E or ΔE^*_ab) to help: CIE76, CIE94, CIEDE2000

The numbers in the names are the years they were published. Presumably later revisions are better so I always use CIE2000.

Things

Objects reflect some, but not all, of the light that falls on them. The biases in the reflectances of different wavelengths are an objects colour reflectance spectrum.

That's not a great name. It's more like a wavelength reflectance mapping. No, that's not a great either.

Just to complicate things, in fluorescent materials some wavelengths get absorbed and then some completely different wavelengths are emitted.

If an object under white light reflects only the yellow wavelengths I'd call it a yellow object.

Not everybody would though. Sigh.

We can make a paint that reflects red light and green light but absorbs all light of other wavelengths. If we paint a car with it and put it out in the sun I'd call it a yellow car. But if we shine cyan light at it, made from a mix of blue and green light, only the green will bounce off it and the blue will be absorbed. It will look completely green. If it's indistinguishable from another green car is it still a yellow car?

I think it is but so many people argue it isn't that I'm not so sure any more. They say that the colour is the perception and what is perceived is green.

That's not without merit.

White light will have metamers that don't include the specific red that would be reflected from our car. Probably. So we might be able to shine a light that's pretty much indistinguishable from white and still see a green car.

Isn't it irritating when the pedants might be right.

I say an object has the colour of what the perception would normally be, if I looked at in full spectrum white light, not under any contrived circumstances.

Clumsy. Reaching.

Remember, a cyan Volvo under a yellow sodium street light against a purple background is going to look…   ugly.

Paint

We have talked a little about mixtures of light, now let's consider paint.

I'm going to talk about paints having colour which isn't technically correct but you know what I mean.

The following assumes a full spectrum white light falling on everything.

Like any object, a paint's colour is determined by what wavelengths it absorbs and what it reflects. When paints are mixed their colour reflectance spectrums are combined.

The resultant colour seems to be determined more by their absorbance characteristics than reflectance. This may be because incident light has the chance of hitting multiple molecules of each ingredient before being reflected. So if any of the ingredients absorb its wavelength then it won't escape.

That seems right - but I don't actually know.

Yellow and blue paints makes green (sometimes)

Most paints reflect a wider range of wavelengths than we'd expect just from looking at them. A yellow paint may well be reflecting many orange and green wavelengths but the sensation is still yellow. Similarly a particular blue paint may also be reflecting some green wavelengths and others. When we mix them only the wavelengths that both paints have in common are reflected. In this case green.

   ROYGBIV
   _oYg___   a wide spectrum yellow
 + ___gBi_   a wide spectrum blue
 = ___g___   green: the only commonly reflected wavelengths
 

Note that other, similar looking, yellow and blue paints may not mix the same.

   ROYGBIV
   __Y____   a narrow spectrum yellow
 + ____B__   a narrow spectrum blue
 = _______   black: no commonly reflected wavelengths
 

This sort of thing can lead to many mixing frustrations. If you want a good, saturated green go and buy one. Use mixing to modify colours a little not try to reach half way across the colour space to get to a completely different one!

Some people like to mix their own blacks but they're never as dark as ivory black or mars black. So don't bother.

   ROYGBIV
   Ro____v   red, but this particular red paint also reflects a little orange and violet
 + __y_B_v   blue, but this particular blue paint also reflects a little yellow and violet
 = ______v   black, but not quite
 

Making black paint is difficult. Most commercial blacks seem to actually be very dark blues. Acrylics painters should check out culturehustle black. It's about as black as you can get.

A colour can have any number of metamers produced by different spectral powder distributions. Different paints of the exact same colour may have very different mixing characteristics. In general you can't tell a paints mixing characteristics by its colour (in any given light).

What we can know is that when we vary the ratio of a mixture of two paints we will produce a continuous path between them through a colour space. The path of mixture will certainly pass through all intermediate values of hue, values and saturation. This gives us a nice handle we can use.

More to follow.

Further notes on paint mixing can be found in Tutorial 1 Part 4 - Colour mixing.