Colour Adjustment - Part 1

Mike McNamee sets off on a journey towards colour perfection

One of the most potent features of digital imaging is the ability to colour correct before output. This strength comes from two quarters, the ability to measure within the image and the ability to correct using numerical values. Both add great precision to the procedure, so much in fact that you can colour correct with impaired colour vision, providing you know the target numbers of various parts of the image.

Colour Blindness are you seeing clearly? Most of us take colour for granted although it has not always been so. Primitive cultures tended not to bother about colour and had a limited vocabulary of colours (that is they did not assign names to specific colours). Colour is in any case a conceptual thing that only really exists in the mind of the viewer. It ceases to exist in a darkened room. Animals see in colour quite differently to humans; dogs for example are limited in their colour discrimination. Our New World cousins have different colour to us Old World types. Forty million years ago primates developed an additional skill that enabled them to differentiate green and red - pretty handy if you are up a tree trying to decide which fruit is ripe and which is not. Colour vision works a bit like your scanner. In the scanner, the detector looks for Red, Green and Blue light and mixes them together to define the colour of the pixel. The eye works in a similar way. "Cones", which are situated close to the central axis of the eye have pigments which are (each) one of three colours. They are usually thought of as red, green and blue although they are, in reality, not too close to those colours, being more accurately described as Yellow-Green, Green-Yellow and Blue. This does not inconvenience us as the brain does clever tricks and makes red look like red etc. The cones are the source of our colour vision and it was a 3rd cone type that formed those 40 million years ago to give us our enhanced 24-bit vision. Dogs have 16 bit vision (2 by 8-bits of colour!). Some poor Aussie nocturnal creatures have poor old 8-bit vision and see in monochrome. Poor vision does not however seem to be suffered by Aussie batsmen.

While the cones are the source of our enhanced colour vision they can also be the downfall of some people. If your genetics are mixed up you could end up with one of your cone sets missing or perhaps with cones that are too similar to each other in their colour detection. The genetic basis of colour vision is what makes men more vulnerable to colour vision defects, 16 times in fact (0.43% of women 8.14% of men to be precise). Girls take X chromosomes from each parent while boys take one of the mother's X's and one of the father's Y's. As the X chromosome carries the pigment making signals for the cones, the boys are disadvantaged.

If you are colour blind or defective there are certain occupations that you are barred from. You won't be much use in a photographic laboratory, an electronics company or as a Police Officer. The first is obvious, the second is because you would mix up all those wires and the third is because the judge will want a better description of the getaway car than "it was a reddy, greeny-grey colour".

Assuming you are not a colour defective i.e. something biologically wrong such as a missing pigment, how do you measure up when judging colour? This is an important question because experience tells us that some people seem to be better at judging colour than others. We spoke to one of the UK's leading experts in this subject, Jennifer Birch of the City University and then got her book from the reference library. In this she says "most people with normal colour vision can acquire superior colour-matching skills if they are sufficiently motivated"

This is encouraging for those who worry that they have poorer colour vision than their peers, but there is not doubt that some people seem to have better judgement than others. If Dr Birch is correct (and who are we to contradict!) then the trick must be to learn from our peers who are judged good at it. Another interesting snippet from Dr Birch is that colour discrimination goes off with time away from the job; you need to break yourself back in after a long holiday or time off work! The problem is two-fold for the colour printer, deciding which way the colour is out and then how to make it right! The advice that filters through seems to be consistent; train yourself up, learn from a more experienced colleague and lastly don't give up. It might also pay to visit the site at

http://www.vieo.com/~kinga/CVDtest.html

and test your colour vision. If you find that you are colour blind take up monochrome photography or try harder!

Here are some general pointers

1. The eye is quite discriminating over colour - watch that you don't end up chasing an impossible standard.

2. The trained eye is even more discriminating, better than many instruments.

3. Practice improves performance.

4. Learning alongside a skilled practitioner is useful.

5. Viewing conditions are important and include - sufficient illumination - correct light colour - correct surrounding colours (don't have jazzy wall paper!) - no glare, especially on your monitor

6. Colours in the green part of the spectrum are hardest to pin down and differentiating cyans is the hardest task.

Key facts:
The eye can discriminate 3 million colours.
There are about 1.6 million pigment colours
A 24-bit CRT screen can display 16.5 million distinct colours
The appearance of a small patch of colour is biased towards the complementary colour of its background e.g. a small patch of yellow
looks reddish on a green background and greenish on a red background.
The eye is sensitive to the size of the colour patch that is viewed. Standard view angles for test patches are 2° and 10°. This concentrates the viewed colour onto the cone rich part of the retina.

We learned a few other little facts on our travels through the world of colour vision defects. For example Viagra has been shown to temporarily impair colour vision - there your mother always said it would make you go blind! Monet painted yellow in his late middle period as his cataracts yellowed his vision and went back to painting blue again after an operation had restored his colour perception to normal.

We can use the list above to guide our behaviour when making colour judgement. In general terms you should rely on your female partner, wife, sister to assist you. As well as two brains being better than one, if you call them in late in the process they will not have been sitting there accumulating eye strain and colour bias, a process known as chromatic adaptation. Also you should take a break from your labours before making a final decision on the colour balance of your image, your first judgement is most likely to be correct and this includes your first impression after a break. Also you should take your time and while this goes against the advice last sentence to some extent it is a fact that only experienced colourists are able to make fast judgements.

Another trick you can employ to assist is to highlight the numerical adjustment values in a Photoshop dialogue box (by double clicking them to make them show blue). Now, when you hold the shift key down and click on the up or down arrows, the values jump by 10 units, producing a rapid shift in colour. If you move without the shift key the values jump 1 digit and you might have adapted to the shift by the time you get 10 clicks in. By jumping in big increments you can do a Goldilock's Porridge adjustment (i.e. - this one is too big; this one is too small; this one is just right!).

A final word before we get into the nuts and bolts of adjusting colour. Do not waste your time chasing perfection in your greys off 6 colour ink jet printers - you will never get there if you are very discriminating. The most recent work by the colour scientists at Derby University has shown that colour discrimination around the neutral is close to 1 delta E. This is also close to the precision of the measuring instruments we use to build profiles. Profiles are intended to harmonise the whole gamut and in practice they tend to sacrifice accuracy around the neutrals in favour of correcting across the whole range. This is particularly so with the 6 ink systems which have the additional complexity of trying to switch from light cyan to deep cyan etc. Experience of showing grey scales to very discriminating photographers has taught the author that some people will be troubled by greys that are incredibly close to neutral. If you are one of these people you should get your enlarger back out of the attic and start mixing chemicals again! In practice your clients are less likely to be bothered by such small errors.

Measuring Colour

This series of articles are planned to be a reasonably complete explanation of the topic and we must therefore set the scene with some care. We have so far explained how colour is perceived now we need to explain how it is measured. We have to understand the relationship between measuring and correcting colour so we know where to put our effort in.

Providing you have either a colorimeter or a spectrophotometer, measuring colour is simple. You simply place the image on a flat surface, plonk the instrument onto the relevant bit of the image and press the measure button. The instrument then sends the colour values to your computer or to its own little screen. It is from here that things get complicated! There are literally dozens of ways of expressing the colour values. This is a bit like currency. A jar of coffee may cost £3.45, $5.18 or 5.48 Euros. The coffee in your trolley does not change but the numbers it costs vary according to how you are going to pay for it. Colour is the same. The colour in the image stays the same but the numbers (currency) you use to describe its value vary according to how you measure it.

Initially we will concentrate on Lab colour. This is the underlying colour system used by Photoshop (regardless of what your image says it is in). Both a CMYK and an RGB image will be worked as Lab values in the background by Photoshop. When it comes to send the colour to your screen, Photoshop works out the voltages to apply to the red, green and blue electron guns at the back of your monitor. However, if you send the image to your printer, Photoshop works out how much cyan, magenta, yellow and black inks to spray onto the page. The fact that these two methods are so radically different is the source of much of the grief you get with imaging and reproduction. In Lab colour, the L value is the lightness, a is the red-greeness and b is the yellowblueness. Hence skin has Lab values of 66L, 14a and 15b. This indicates that the brightness is 2/3 of the way towards fully bright white (66% is almost 2/3 of 100%) and that the other predominant colours are red and yellow. Negative values of a indicate greens, negative values of b indicate blues. Hence a colour with values of zero for both a and b is a neutral grey, of a density determined by its Lightness value.

In order to understand colour better, scientists plot the colour on a graph. This is like a map and just as you can measure how far two places are apart on a map, you can measure how much two colours are different on a graph. Just as you can look at a map and say Denbigh is west of Liverpool you can say that magenta is more towards a red than a cyan.

Crucially, the eye's ability to judge colour varies depending upon what the colour is. You have a hard time deciding if a deep black has a blue bias or a green bias for example. It is easier to spot the difference between two reds than between two deep blues if they are different by the same amount (i.e. the same distance apart on the graph). Crucially, the eye is most sensitive when spotting the difference between light grey values. This is why you are rarely satisfied with your grey scale off a 6- colour printer. Not only does the printer have a hard time getting the colour right, your eye spots the errors with great ease!

Right:The Lab Plot. Pure Yellow is at the top, pure blue at the bottom. Green is to the left and red to the right. Also superimposed are the discrimination elipses. Note that they are small in the centre (easy to discriminate), large towards the pure colours and largest of all in the deep blues. The shape of the elipses indicates that the eye is more sensitive to changes in hue (colour) than to saturation.

The SWPP 2008 Convention was an outstanding success,
we have 193 days to get ready for the 2009 convention - which starts on January 14, 2009

Photo Quote: Above all, it's hard learning to live with vivid mental images of scenes I cared for and failed to photograph. It is the edgy existence within me of these unmade images that is the only assurance that the best photographs are yet to be made. - Sam Abell