1.2 Basics of Colour

This chapter is all about colour and the human eye. In particular we focus on the three parts of a particular 'colour event' - light, an object that reflects that light, and a sensor that receives and interperets that light.

We also introduce a few basic concepts - gamut, metamerism and linearity/non-linearity.

Colour Basics

Introduction

This chapter is about laying the groundwork.  We’ll get to the really good stuff in a week or two, but there’s a few concepts and ideas we have to cover first.  Crawl before you can stagger home, as they say. 

We’re going to cover a bit about the human eye and why all this is so hard, a bit about what colour actually is, and then some really geeky stuff about how computers think about images.  All this will lead directly into how to control that very fluid thing that is digital colour.

The Incredible Human Eye and its Fickle Nature

The human eye is an awesomely sensitive measuring instrument, capable of seeing the universe in ways your camera can only dream of.    Printing, indeed photography, is a con, a great big trick.  It’s all about using a pathetically limited medium, a single, generally flat, piece of paper, to fool the human eye into thinking that it is seeing something vaguely related to rich and wonderful reality. 

Our challenge is to use the very limited tools we have available – contrast, colour, that sort of thing – to create a miniature world, capable of evoking emotion/stimulating thought/firing of the visual pleasure synapses. 

What the world presents to the eye, and what they eye can take in, are so far beyond anything realisable on paper that it is amazing we can get away with this trick at all.

Lets look at contrast as an example.  Imagine yourself on a beautiful tropical beach.  Your eye, quite comfortably, will see detail in fluffy white clouds, and in the deepest shadows of the palm trees.  I don’t know the exact figure, but it is usually quoted as being over 100000:1 – that means you can comfortably look at a scene and see detail in the brightest object, where the brightest object is at least one hundred thousand times brighter than the dimmest object you can still see the detail of!   Your camera can’t come close to capturing anything like that brightness range.  Prints, even made with the very best printing systems, can’t achieve anything like that range of contrast – not even close. 

Looking at colour, we’ve got the same problem.  The human eye can see a very broad range of colours indeed.  Although, by animal standards, we’re pretty feeble.  You might think a crow is a black bird but put it under a black (UV) light and you’ll see that it isn’t – it’s actually very colourful (to other birds, anyway).  This is, perhaps surprisingly, quite relevant, because it makes us very conscious that when dealing with colour, we only care about what is commonly referred to as the visible part of the electromagnetic spectrum – that is, the only colour that really counts is the colour we humans can see. 

From our high school science lessons, we’re aware (hopefully) that off one end of the spectrum is Ultra-Violet and off the other is Infra-Red, but we all know that ROYGBIV is all that really counts to us (ROYGBIV = Red Orange Yellow Green Blue Indigo Violet).  But already this should be giving you the feeling that colour is all about the observer – colour doesn’t exist unless its seen (if a tree falls in the forest and nobody hears it, are its leaves still green?). 

To get back to the topic, the human eye can see a great deal of colour.  And of course, just like with contrast, we can’t print anything like that amount of colour.  We work within a pathetically small subset of colours the eye can see.  You’ll see precisely how small soon.

What The Eye Wants

The eye, being such a sophisticated sensor, capable of such amazing things, craves these very things.  Like any addict, it craves a bigger hit.  It’s just like when you’re listening to music, and over the evening, you turn it up, then up some more, then eventually, around 5 o’clock in the morning, the neighbours are thumping on your walls and you’ve got yourselves a little problem with the landlord. 

The rods and cones in your eye, those things that actually detect light and send the signal to your brain, get saturated with signal.  So as you look at an image for an extended period, and Photoshop it from mere snapshot to work of Art, be aware that your eye is going to start lying to you.  It’s going to tell you that your image needs a little more contrast.  Maybe a little more saturation.  And then a bit later, hmmm, maybe it needs a bit more punch!  I know, I’ll boost the contrast some more. And a little colour.   Pretty soon, you’re beautiful fine art portrait looks like a Hunter S Thompson novel.

The point here is that the eye needs breaks – when you’re working on images, what you see is not the same as what someone looking at your image for the first time sees.  It’s not the same as what you will see if you go away, have a coffee, and come back to the image.  So take regular breaks and give your eyes a chance to rest.

What Is Colour?

Seems obvious, really.  ROYGBIV, right?

Well, that’s only one part of the story.  Colour is a remarkably fluid thing and pinning it down is next to impossible.  Long, dull books have been written on colour theory.  We’re just going to cover the basics, so you can see why all this stuff is so hard, and so you can understand the tools you’re going to be using.

Colour involves three things – a light source, an object to reflect that light, and a sensor to receive and interpret that light.  Think the sun, the ocean, and your eye.  Specifically, the visibly wavelengths of light are bounced off the object and stimulate the rods and cones in our retina to send signals to your brain.  Obviously, your perception of that colour will change if you vary any one of those three things – that is, if you use the moon instead of the sun, the sandy beach instead of the ocean, or you’re a little red/green colourblind. 

So pinning down colour can be hard – no object IS a particular colour, it just appears to be that colour under certain conditions.  There are no absolutes in colour – unless you control all three of those variables, you can never absolutely say what the colour of something will be and how it will be perceived by others.

So we’re going to talk about the basics of the three things involved in colour:

Light

Light can be thought of as a wave, and we can see wavelengths between 380 and 700 nm. This is the visible spectrum, and what we commonly refer to as, simply, light.  From 380 to 700nm, we have named the colours, and it looks kind of like this:

  UV 380 nm Violet Indigo Green Blue Yellow Orange Red 700 nm IR
  1

While IR and UV are not generally visible, they do have some impact on fine printing, so don’t forget about them completely.

(IR is most relevant and the capture end as digital sensors are very sensitive to IR, hence the IR filter in most cameras, and UV is most relevant in printing because they show up in fluorescent optical brighteners).

The light we see is typically made up of many wavelengths, that is, our eyes simultaneously see and mix multiple wavelengths in day to day use.

White is what we get when light is made up of equal amounts of all the wavelengths of the visible spectrum.   But
think about a white piece of paper for a moment.  The eye is pretty amazing, almost no matter what the light source, whether it’s the sun or tungsten lights or glow worms, our eyes will pretty much always look at a piece of paper under any light source and go ‘that’s white’.  This is called colour constancy.  Our eye is great at ignoring, or coping with, different light sources.  It’s one of the things that makes colour really hard.  Film, scanners etc, are not – that’s why we have different films for different types of light sources, tungsten, daylight etc. 

So we can see there’s no true definition of white, no such thing as absolute white.  So we use a term known as colour temperature to talk about the actual colour of white light that we mean.  You should know all this from first year basics, but it’s worth revising now as it will have a pretty big effect on things later.

Objects

Objects reflect light.  Which particular wavelengths they reflect determine the colour of the object.  They can of course, only reflect those wavelengths that are falling on them in the first place.  That’s why flourescent tubes can make things look funny, because they simply don’t output certain wavelengths of light, so objects can’t reflect them, so they look odd under flouros. 

That’s about all you really need to know, except it is worth considering one other thing…

Fluorescence is a property of objects whereby they change one type of wavelength into another – and it is most noticeable in really, really white things.  Like your teeth, if you brush them.  You know what teeth look like in a night club with those dodgy UV lights?  Brighter than white.  Well, the same stuff they use in toothpaste is also used in detergents for clothes, and more importantly, papers, to make them appear really white.  They’re called Optical Brighteners and some people get pretty upset about them.  Probably unnecessarily, but the jury is still out on that to a certain degree.  We’ll talk about them later.

Fluorescence is often important in colour management and digital printing because it can make things unpredictable, and that’s a big problem in colour. 

Sensors

Sensors – your eye is one, your camera another, and scanners are yet another.  In some ways, they’re all very similar, in other ways, very different indeed.  The Eye is *very* flexible in its response, whereas cameras and scanners are very much fixed in theirs. 

The eye, your camera, and scanners are all Red Green Blue sensors.  This is really important – it’s the basis of all colour reproduction.  Our eyes and these machines  mix these three colours to see all the rest.  On paper, we subtract reflectivity from the paper using the opposites of these colours (Cyan, Magenta, Yellow), to produce all visible colours.  That is, in any colour system, we can use just three primary colours to make all the others, which is pretty amazing.  How that works is pretty complicated, but it will do just to know that it does, generally, work.

Red, Green and Blue are known as the additive primaries – because we add them together to get all the colours. 

RGB is the language of light.

Cyan Magenta and Yellow are known as the subtractive primaries, because we use these colours to subtract reflectivity from white (i.e. paper) to get all the colour. 

CMY is the language of ink.

Some of you are bound to be thinking, right about now (if you’re still awake) it is CMYK not just CMY.  So what is that K all about?  The K stands for black, and it is added into the ink mix for two main reasons – one, to get better, purer blacks, and two, because it is cheap.  And for most printers, cheap works.  When we talk about CMYK later, we’ll go into more details.

It’s worth noting that there’s a pretty fundamental difference in the what’s going on in light based systems versus ink based systems.  In light systems, you get more saturation by adding more light together, so your saturated colours tend to at the higher end of the brightness scale.  With ink, you make more saturation by adding more ink (and therefore subtracting from the paper’s reflectivity), so you tend to get your more saturated colours at the lower end of the brightness scale.

The two sets of colours (RGB and CMY) are complimentary (or opposites) – here’s the digital colour star:

4

As you can see, The digital colour star shows you the way these things work – if you’re trying to get rid of blue, you need to add yellow (or reduce both cyan and magenta). There’s no such thing as a yellowy blue!

It’s worth drawing yourself one of these and keeping it handy for a while, it really helps when you’re trying to make really subtle changes to colour to have a good grip on how the colours all relate with each other.

How you might want to think about colour…

The upshot of all this is you need to think start thinking of colour as a fluid thing, dependent very much on the properties of the original light source, the object you’re looking at, and what you’re using to look at it (i.e. the sensor involved).

Given there are so many variables in colour, it’s hard to talk sensibly about colour without defining a few thing.  So we have to define a absolutes, or reference points, so that we can get to grips with all this.

The most important of these are the reference light sources.  You may have heard of these – in particular you may have come across D50 and D65.

D50 is a light source with a whitepoint of 5000 kelvin.
D65 is a light source with a whitepoint of 6500 kelvin.

They both have a bit more to them than that (notably they have defined spectral output curves which define exactly how much of each given wavelength of light they emit).

Once you have a precisely defined light source, you have nailed one of the three variables to the wall.  In theory anyway.  In practice, you will almost never come across true D50 lighting.  Fortunately, there are a lot of 5000K balanced lights out there that come acceptably close.

The next thing we nail to the wall is the range of colour the human eye can see – the only colour that is really relevant.  This has been measured for us, and is called the 2˚ Standard Observer.  This results in a particular colour space called CIELAB, which is a description of colour as the human eye sees it – roughly.  More on this later.

We’ve now standardised two of the three variables – the light sources and the observer.  All that is left now are the objects, i.e. prints.

Well, the objects we create are ink on paper, and how skillfully we put it there is what determines how good a printer we are.  Of course, unless we fancy grinding our own pigments, we have to use the inks we can easily buy to produce our prints, so the trick is to really understand the behaviour of those inks and the printers using them, to create prints which look good under standard light sources to standard observers!

In the real world, the reality is that we don’t normally have the standard light source OR the standard observer available to us, so in reality we’ll be working within acceptable tolerances, and it all works pretty well in the end.  But you have to hang your hat somewhere, and for colour, the standards above are the commonly used hat racks.

A few more concepts we’re going to need…

There’s a few other bits and bobs we all have to be on the same page about before we launch into the business end of colour management.

Gamut

Gamut is just a word that means the total range of colours of something.  The gamut of the eye are all the colours an eye can see.  The gamut of a printer is all the colours that printer can reproduce (which is determined by the inks the printer is using, the paper/coating the inks are being put on, and the driver technology that is actually determining how that ink is being put on the paper).

Metamerism

A true definition of metamerism is quite complex, but metamerism in practice is the phenomenon whereby something you print changes colour unpredictably under different light sources. 

This is most relevant to black and white printing, and is still a very common problem.  Basically, say you can achieve nice, neutral greys out of an inkjet printer, or so you think.  Take your nice neutral great print outside and it suddenly changes colour, taking on a unpleasant cyan or magenta cast.  This is metamerism in practice, and in many cases, it is an unsolvable problem without pretty fundamentally changing your inks. 

Linear and Non-Linear

Here we’re talking about maths, and input output systems in particular.  As an example of an input output type of system, think of curves, which are a classic input output system.  As you move the curve in Photoshop, say dragging it up in the middle, you are changing the relationships between tones from your original image to the newly created, more bright image.

There are two fundamental ways to change things in images – either I can change ALL pixels by the same amount (e.g. move all pixels up 10 levels brighter) – a linear change because it is constant across the entire data set,  or I can change some of the pixels one way, and other in different ways (eg. an S-curve in Photoshop, where I am lowering the values of shadows by 10 pixels and raising the values of highlights by 10 pixels – a non-linear change.

That’s one, simplified, way of thinking of linear vs non-linear.  It’s not the mathematical definition but it works.

Linear things are easy – If you increase the input end by 10%, the output end goes up 10%.  But many things in real life are non-linear.  For example, when you double the volume on your stereo in terms of decibels, things don’t sound twice as loud to the human ear. Louder yes, but not twice as loud.  So many things like this are said to have a non-linear relationship.  Brightness is one of the key non-linear relationships that we’ll talk about.  If twice as many photons of light hit your eye-balls, your perceived level of brightness is NOT twice as much.