
| Table of contents |
This article tries to set out what I think is the bare minimum that any digital photographer needs to understand about colour management if he or she wants to take their work at all seriously. Colour management is a very complex subject and many thick (and largely indecipherable) books have been written on it. So what follows is the merest scratching of the surface, but I think it’s enough to be going on with for most of us most of the time.
Before reading on, please imagine I have asked you this question, “How much do you know about colour management?” If the answer is, “Nothing – what’s colour management anyway?”, then this article is for you. If the answer is, “A bit, but I don’t feel very confident about it”, then you may get some revision benefit from a quick scan through what follows. If, however, the answer is, “Probably a lot more than you, Steve”, then please don’t waste the next five minutes of your life reading this. Waste them doing something else!
Objects are coloured either because they emit coloured light themselves, or because they reflect light coming from somewhere else. We all know from school physics that light that appears white is really a mixture of light of many different colours, and that white light can be split into its various colour components to form a rainbow, or spectrum, of colours from violet at one end to red at the other. When we think of light being emitted, we use the additive model in which light of different colours is mixed (added) together to form a huge range of colours depending on the relative quantities of light from different parts of the spectrum. When we think of light being reflected, we use the subtractive model in which various components of white light are absorbed by the object in question, so that what is reflected back is again a mixture of light from different parts of the spectrum, but this time white minus whatever has been absorbed.
In photography we need to understand both these models. When we take a picture of a scene, it is certain to contain reflected light, but may well also contain emitted light, perhaps a lamp or a neon sign, for example. In digital photography, we work on images displayed on a computer monitor, and this is emitted light. Frequently we want to print our images, and a print is obviously an example of reflected light.
Although white light when split into a spectrum is traditionally thought of as being made up of 7 colours, we find that every colour can in fact be produced from the adding together of different proportions of 3 colours only: red, green and blue. When all three colours are mixed in equal proportions, the light is perceived as white, or as grey if only a little of each colour is emitted. When no light at all is emitted of any colour, then the result is black. It is an extraordinary but true fact that every nuance of colour can be produced simply by varying the proportions and the absolute levels of emission of these three colours. For convenience we refer to this colour model as R(ed)G(reen)B(lue). This is the model used by computer monitors. It is also used by digital cameras, and it’s useful to think of the image captured by a digital sensor as being like a colour slide with a red, a green and a blue layer.
If an object is illuminated by white light containing a continuous spectrum, then its colour will be determined by what colours it absorbs, and therefore by the mixture of light that is reflected back, which will be what’s left once the absorbed light is removed. In the same way that all emitted colours can be created from the relative proportions of red, green and blue light, so every colour can be created from a mixture of dyes or inks that absorb red, green or blue light. If all the green light is absorbed, what’s left is red and blue, which together make magenta. If all the red light is absorbed, what’s left is green and blue, which together make cyan. If all the blue light is absorbed, what’s left is green and red which together make yellow. If all the light is absorbed, what’s left is nothing, and nothing looks black. (Well, some things do look black, but you know what I mean!) This model is therefore called C(yan)M(agenta)Y(ellow)K. We use K rather than B for black to avoid confusion with B for blue. This is the model used by printers.
Digital photography is a continuous process of creating colours by the application of these two models. In the camera, incoming light is split into RGB components, and a monochrome image of each component is recorded on the sensor. The firmware of the camera controls the processing and analysis of this data, and produces the three monochrome versions by recording the RGB and brightness information for each pixel in the image. Then, after we download the file to the computer, the computer software and hardware interprets this data and sends it to the video card, from where the screen firmware re-interprets it for display via the RGB phosphors on a CRT screen, or the LCD elements for a flat screen. We manipulate the image in an editing program, and then save a file which is in turn sent to a printer which must convert the RGB data into CMYK data, and then represent this via the mixture of inks or dyes that it deposits on the paper. When you consider how complex each of these steps is, it’s a miracle that what we get out of the printer bears even a slight resemblance to the scene we saw when we took the picture. It works because all the software writers and hardware manufacturers work to a set of industry standards, so that within limits all the conversions and data interpretations work to the same rules. But only within limits.
For most of us, the slight discrepancies that are introduced by the limits of accuracy only really cause us trouble when we compare what we see on our computer screen with what comes out of our printer. Our vision is not at all good at absolute colour, but it’s very good at relative colour. If I show you a swatch of material today with a range of colours in it, and then show you another swatch next week and tell you that the colours in this one are identical to the one I showed you last week, it’s highly unlikely that you’ll be able to argue with me unless the colours are completely different. But if I show you both swatches together, you’ll be able to distinguish even the most subtle differences. So if what’s displayed on your monitor is even slightly different from what your print looks like it’s immediately, and usually also painfully and annoyingly, obvious. So for most of us the most important thing is to be able to control the colours on our screen and the colours output by our printer. Unless you are a professional photographer taking pictures of fashion or works of art, for example, it’s not so important whether or not the final colour on the print is exactly the same as the colour of the object you photographed. Indeed, most times you won’t even be aware of these absolute differences anyway for the reasons I explained above. This is fortunate, because for the photographer for whom absolute colour faithfulness is critical things are very much more complicated. Such a photographer must be concerned with the colour temperature and spectrum of the lighting which illuminated the scene, with profiling the response curve of the camera, and with the minutiae of the particular printing technology or other display media (perhaps projection, for example) that’s to be used as well as the profile of the computer monitor. Oh, well, that’s their problem, not ours!
A monitor’s representation of an image is controlled by the interaction of 4 key things. First, the black point: black must be black, not a murky grey, or a very dark blue. Second, the white point: white must be white, not a very pale pink or a dirty grey. Third is the contrast curve: evenly spaced steps in the brightness of the image must be displayed by even and equal steps in the luminance of the screen. Finally, the mixing of RGB colours must be accurately recreated so if a colour should be 40% red, 30% green and 30% blue it’s no good displaying 50% red, 20% green and 30% blue. Monitors are notoriously inaccurate, and they are also variable. Their display changes over time, so what’s right today will very likely be wrong by next month. There’s only one way to deal with the slippery world of monitors, and that’s with calibration tools. These consist of a colorimeter placed over the screen, and software that calculates the difference between what the video card “told” the screen to display, and what the colorimeter discovers the screen actually did display. The software then creates a compensating filter that alters the video card’s instructions in such a way that the screen does in fact display what the video card “wanted”. This filter is stored in a file called an ICC profile, and you must have this profile loaded into the video card every time you boot up.
When you add a printer to your computer’s configuration, the printer manufacturer will provide a driver that translates the RGB output from your image file into the CMYK format it needs to use to print. Here again there’s plenty of room for error, but fortunately printer manufacturers usually provide ICC profiles which are specific to the particular printer and paper combination. So, one profile might be for glossy paper whilst another is for matte. There are tools for calibrating printers which will produce a more accurate ICC profile for your individual example of the printer model in combination with your specific paper stock and the actual inks or dyes you are using. However, if you stick to the combinations of printer and paper stock for which your manufacturer has provided ICC profiles, then in my experience that is adequate.
The key issue is this: if I spend time carefully adjusting my image for contrast and dynamic range, correcting any colour casts or other defects, I will end up with an image on my screen which is the best that I can make it. If I haven’t calibrated the screen, nor installed an ICC profile for my combination of printer and paper, I can have no idea whether my careful work will be faithfully carried through into the print. I might as well not bother at all.
Unless you are prepared to invest in a monitor calibration system, and to download the correct printer profiles, don’t waste time post-processing in Photoshop or other editing software. It’s like driving with your eyes shut. So this is my minimum colour management setup:
You will find that you can then work on your images in the confidence that what you do on the screen will be faithfully reflected in how the image prints.
Steve Brown lives and works in London, and has moved from film to digital over the last two years. His main photographic interests are architecture and the French countryside. More of his images can be seen in his TLF gallery (http://www.thelensflare.com/profile.php/sgbrown) and on www.hautevienne.moonfruit.com.
