In order to answer this we need to understand a little about the human visual systems and the steps involved in printing a picture.

Why is it that only three basic colours (red, blue and green) can generate an apparently infinite number of colours in the case of a TV screen? Similarly, why are most pictures printed with only three or four colours (cyan, yellow, magenta and sometimes black)? The answer lies in the human visual system. In the human retina there are two types of light-sensitive cell, rods and cones. Rods are used for colourless vision at low levels of light such as moonlight. Cones on the other hand, provide the sensation of colour. In fact there are three types of cone, each sensitive to a different part of the visual spectrum. Although scientists were aware of the different cells in the retina it was only in the 1940's and 1950's that the three types of cone were identified and found to be particularly sensitive to light at approximately 445nm, 545nm and 600nm, corresponding to reddish, greenish and bluish light respectively. Thus, as the eye is particularly sensitive in these three parts of the spectrum, it can be fooled into seeing a complete range of colours by varying the amount of red, green and blue light entering the eye. This happens in a television screen where each spot on the screen can emit varying amounts of the three primaries.

In the case of a printed picture it is a little more complicated. Light from a source such as a desk lamp or the sun is reflected off the printed page into the eye. Natural objects such as flowers or birds contain pigments of different colours. There are an infinite number of pigments and the colour of each flower is slightly different causing a certain variability that provides colour richness. This is readily observable by looking at a field and noticing the different shades of green. Artists try to capture this richness by hand mixing different paints. However, when printing from a computer, it is impractical to have an unlimited range of inks. But as we know the eye is particularly sensitive to certain colours and we can exploit this feature by choosing inks that stimulate the red, green and blue cones in the eye. In this case, however, we need to filter out unwanted light and unlike a television, which emits light, every time we use ink the reflected light is darkened as the ink absorbs light. In order to cover the greatest range of colours and to minimise the amount of ink we use the complementary colours of red, green and blue namely cyan, magenta and yellow. Cyan absorbs red light, magenta absorbs green light and yellow absorbs blue light, thus giving us a way to vary the amount of light to which the cones are sensitive. Black ink is used to increase contrast, compensate for ink impurities and reduce the overall amount of ink required. The leads to the subtractive colour circles that we all remember from school days:

Figure 1 - Cyan, Magenta and Yellow subtractive primaries, combined to produce Red, Green, Blue and Black.

The science of colour is particularly important today as colour printing is moving from the exclusive domain of the colour professional (who combined art with science) to the home by way of colour digital cameras, colour scanners and colour printers. Everyone expects to get the same results as a professional printer for their home photos but as anyone who has tried digitally printing pictures knows, consistent results remain elusive.

Several problems need to be overcome:
· Communication of colours - in a computer colours are typically stored as triplets of red, green and blue values i.e. an additive colour space. Typically each colour has a range of 0 to 255. Common colours are:

In most cases however, these values are relative to the input device, thus scanning the same picture on two different digital scanners may produce quite different results. Calibrated colour spaces do exist but their use is mainly limited to expensive systems.

· Digital half-toning - most pictures are printed from only three or four inks (cyan, magenta, yellow and black) and in order to produce a wide range of colours we need to exploit the eyes ability to 'merge' closely spaced dots of different colours into a single composite colour. The smaller the dots the more convincing the composite colour appears to the eye. When an image is printed from a computer, the printer and the computer interact and between them they decide where to place each individual colour dot. Computers use algorithms (recipes) which they slavishly follow, to decide where to place the dots. This sometimes results in annoying repetitive patterns that are clearly visible, particularly when there is a large area of exactly the same colour to be printed.

· Colour Half-toning (see Figure 2) - half-toning has been used for monochrome printing for may years but unfortunately printing in colour is more than just separating out the three or four colour planes (which in itself is not a trivial problem) and using a halftone algorithm on each tone separately. Colour dots printed on top of each other react differently to light when compared to dots printed next to each other. This means that the colour half-toning algorithm must either keep track each printed dot and its colour or use probabilities to keep the colour deviation in check.

We can now answer the original question - first we have wide differences in the quality of scanners and cameras especially in the range of colours they can see and their ability to communicate them effectively.
Secondly, we have many complicated algorithms at work to place each dot of colour on the printed page. Usually the more accurate the algorithm the longer it takes hence a compromise is made between quality and speed.
Thirdly, we have limitation in the inks that are typically found in computer printers. They are selected for a number of properties including cost, stability and viscosity. This results in colour impurities that frequently limit the range of colours that can be printed.
Finally, there are psychological aspects to print quality, we are all used to seeing very high quality (and often artificial, highly saturated) images in magazines which seem to exaggerate the defects in our own printed pictures.

Work is ongoing to accurately characterise input devices such as digital cameras and scanners; output devices such as printers and monitors; and to improve the computer algorithms that map one to the other. Eventually, this will enable pictures taken with any camera to be sent from one corner of the globe to another and faithfully reproduced on any colour printer.

Figure 2: Example of colour separation using exaggerated screens of 10 dots per inch. Clockwise (a) Original picture (b) Cyan screen at 60o (c) Yellow screen at 90o (d) Magenta screen at 450.

---