A color space is a set of colors also called gamut. The gamut of a monitor represents all the colors it is able to display. The gamut of a printer represents all the colors it is able to print. There are two types of color spaces: the ones relative to a device (and we're talking about ICC profile then) and the ones created by engineers so that they don't depend on a device and among which the ost famous are sRGB, Adobe RGB 98 or also ProPhoto. The largest, representing the set of colors that a human being can see, is called L*a*b* space. All the others are smaller because technical limits keep devices from reproducing so many colors, for now at least. Then why did we invent so many?
Key points if you are a beginner ...
Here are the key points to remember on color spaces. The rest of this page will be dedicated to those who want to go into more details.
Note ! The concept of color space is not simple and is primarily dedicated to advanced users, or those who really want to know what happens in Photoshop when retouching colors. All softwares, like Photoshop or Lightroom, take care of everything, they choose the right options and you will have nothing to do. So do not panic !
What is a color space ? A color space is a set of colors. This is called a gamut. All colors a human being can see is called L*a*b* color space. This is the largest color space. There are many other, smaller, that of YOUR camera, your monitor, your printer ... In fact, your smartphone, your camera do not succeed to photograph all the colors that you are able to see. But do not worry, they see the major ones and it is quite enough already. The rest is for the purists or really specific cases.
More importantly, a camera, a monitor or a printer see, display or print colors with defects (often small but real. These are technical limitations). Color spaces of those devices are limited and imperfect. It's the ones, color management will try to correct.
Finally, there are others, specific, because they do not depend on a device but are virtual, ideal and perfect. You may have heard of sRGB. This is the one of your images on Internet. But we need one more thing to correct the defects of our devices ...
The concept of color space - or colorimetric space - is very important in color management. The most important is your images'. The famous Photoshop workspace is often presented as being very important as well but it only is in a particular case: when your image doesn't have an ICC profile at the opening in Photoshop! Let's have a look at that now ...
A color space represents a plurality of colors, visible for a human being or a device. This set is also called gamut. There are several color models or color spaces : RGB, CMYK, L*a*b* orCIE xyz. It is said that some models are dependent upon devices and others that they are independent. Finally, some are called workspaces. Why ? Which is the one to choose in photo editing software? In your camera? From which one can we represent numerically a single color, that is perceived by a standard human eye, to answer the questions above?
The starting point: CIE xyz & L*a*b* spaces
Colors come in about eight million different shades for a "standard" eye (standard but still powerful!). CIE and mathematicians needed to mathematically represent a three-dimensional graph (i.e. a volume), two of which appear in the figure below. The figures below might look familiar to you: that's the chromaticity diagram and the L*a*b* space. They represent all the colors that can be seen with a human eye and will therefore be put to use all the time thereafter.
Here are two different ways to represent the same thing: the colors a human being is able to see. Note: in the diagram on the left, the representation is more accurate to the "way" the eye sees colors since the green surface is twice larger than the red and blue as seen in the previous page.
But why are colors represented in so many different ways?
The aim of color management is, to make it short, to keep the same colors from one device to another: I want to see the color I shot first on my monitor and then on my print! But everything gets complicated then since these three devices don't "see" colors the same way. They all have different defects and they don't all "see" the same most saturated colors. So each device has its own color space, also called ICC profile. To share the same color from one device to the other, it is thus necessary to alter slightly the RGB values to take into account the defects, or characteristics of each device. This is called a conversion. Behind this conversion lies a complex mathematical formula. But mathematicians have invented several mathematical formulas because none, single and unique, could fulfill all the complex specifications of human vision. In some circumstances, they use one and in other circumstances another. Each mathematical formula has its "color space" and its characteristics (assets/drawbacks).
Note that not all these mathematical formulas needed to use RGB values.
Visual analysis of the figures above: in the chromaticity diagram, green is over-represented compared to red and blue, but really, the eye is twice as sensitive to this color. It is the model that comes closest to the human eye. In the L*a*b* space, a color equals one point. Depending on the mathematical "work" to do, mathematicians rather use one color space or another.
CIE spaces are the largest color spaces, as they represent all the colors we can see but they represent colors differently. Since they do not depend on a device but really on the modeling of the human eye, they are said to be independent.
In the CIE xyz space, each color is represented by three coordinates: XYZ (Z as depth, so the brightness of each color) as shown in the figure above. These values, admit it, don't speak for themselves but are very convenient to use in a mathematical formula!
The L*a*b* color space, as for itself, is an absolute colorimetric space, as CIE XYZ, but represented differently. This is also this absolute color space that is used by Photoshop and Lightroom, by the way, as a basis for color management in these programs. It is equivalent to the cash desk to make an analogy with the different currencies of all countries. This is the same as for money changing euros into dollars through an intermediary, the change rate, in order to keep the same absolute value of money in your wallet. Even if you had 100 something in a country to buy gold, you would have 30 in another country, but you could also buy the same amount of gold. 100 ≠ 30 in numeric value, but between the two countries, it has the same value in gold. Here the gold would be the equivalent of the L*a*b* space value, 100 and 30 being RGB values.
Instead of describing a color with XYZ values, the L*a*b* space describes them with L, a and b values. L for the brightness level (0 to 100), a for colors from red to green and b for colors from blue to yellow (-128 to +127). Last point, these color spaces are neutral because they have been defined by a human brain.
And finally, since it is not so convenient to work in L*a*b* values, engineers invented spaces based on RGB values, much easier to use! Then why isn't there a one and only RGB space?
Other color spaces: RGB / CMYK spaces
As you probably noticed, RGB or CMYK values haven't been mentioned any time with these two giant color spaces. However, us photographers almost always work with these spaces, just like our eye. L*a*b* and CIE XYZ spaces are not convenient enough when working, for example, in Photoshop,so other color spaces have been invented based on RGB or CMYK color spaces. Here, colors are represented by RGB values and not esoteric XYZ or L*a*b* values that don't speak for themselves at all! Adobe RGB 98, sRGB, Don RGB, ECI RGB or Prophoto spaces or even my monitor's or my printer's space are RGB color spaces. Here you might want to ask: then why didn't they invent a one and only RGB space as broad ar L*a*b* because what can do more can do less!
First of all for historical reasons. When color management was implemented in personal computer in the nineties, their power was absolutely not equivalent to their power nowadays. And treat as much information per pixel was completely insane considering processors and graphic cards back then!
Secondly, for technical reasons: was it really useful, and is it now? Well there's no camera, monitor or printer that is able to reproduce as many colors as a human eye nowadays, so there's no point in working with eight million colors all the time... So come what may, there was no reason to invent an RGB space as broad as the L*a*b* space. Then why didn't they invent a one and only RGB space, but smaller? Quite simply, to fix numerous problems and because each device has its own, bigger or smaller.
Another important point is conversion. When you send the RGB values of a photo to a printer, they're modified to take its characteristics into account (or said less diplomatically, its defects). This change is called a conversion. And when the conversion is performed from one very wide color space (a camera for instance) to a much smaller color space (a printer's), there are important risks to see tone breaks in the final print. It is thus better to work in an RGB space that is a bit bigger than the final space. We'll also see that later.
It is clearly visible that even the wide DonRGB is much smaller than the L*a*b* space. This one is today quite close to the gamut of an inkjet printer on glossy paper.
Dependent / independent of a device?
Everything would be perfect in the world of color management if all monitors and printers could reproduce or display all the colors that a human being can distinguish and even more without introducing defects! But unfortunately, that is not technically possible or would be too expensive...
The set of colors that a device can reproduce - its color space, also called gamut - is always smaller than the L*a*b* space, due to technical limitations. When a device is given to see, display, print a one color, it is impossible to know, as long as it has not been calibrated and characterized (i.e. compared to a reference, using a specific device or as long as it does not have an ICC profile), what it actually perceived, and what it can display or print. The limits of each device's color space, that can only be known through its calibration, are represented in the figure below, and are the result of:
monitors: min/max brightness & maximum saturation of each pixel (so directly linked to the quality of the RGB filter used);
printers: black color depth / paper whiteness & saturation of the dyes or ink pigments.
In the figure on the right, the L*a*b* color space is represented by the colored square and the limits of other spaces by continuous lines of different colors. The green line represents the gamut of a printer on glossy paper and the red line the gamut of a Graphic Art monitor, close to sRGB. Both are significantly smaller than the L*a*b* space and it is clearly visible that the printer is capable of printing cyan/green that the monitor is unable to display! In terms of surface they are about four times smaller than the L*a*b* space, what means about two million colors.
But there is another characteristic of devices color spaces that can not be seen here: their defects.If you take the XXX RGB value, hence three times the same, the color won't be represented at the intersection of the lines -a+a/-b+b but very slightly apart. Grey values are in fact slightly colored: it is their defect that these two lines (green and red) don't show in this figure. And each device has its own, slightly different every time. Please note, by the way, that these differences have decreased a lot from 2000 to 2016.
So, on paper and in this representation, the difference related to the defects of each device regarding the color spaces DonRGB, sRGB, Adobe RGB 98 doesn't appear and yet devices color spaces are said dependent. Please note, finally, that they're always in RGB.
Color spaces relative to a device
A color space relative to a device describes :
all the colors that it can acquire (scanner, digital camera...), display (monitor), or reproduce (printer) compared to L*a*b* colors. This color space is also called the GAMUT.
All its "defects" or colorimetric characteristics more accurately. During the calibration and characterization process of a printer, for example, the calibration device measures all colors it is able to print, but also how it does it. Those are called its characteristics. In other words, does the printer print colors right! For instance, it is asked to print an L*a*b* color and it prints a slightly different one. This difference and the way to fix it are contained in the ICC profile.
To do so, it is given to print reference colors - a reference chart - and it is going to print different ones!! This information - differences between the color that should be printed and the color that is actually printed - is placed in a special file: the ICC profile, that we'll sutdy in details in the next page. Moreover, no printer is perfect and no calibration device either. Even though they are insignificant, errors in the ICC profile will be real. However, with a good equipment, it is obvious that perfection is so close that we can easily consider the colors are fully corrected and neutralized.
Color spaces not relative to a device
In a space that is not relative to a device, the color space is the result of a calculation rather than a measurement. It is broader or smaller but neutral. The same percentage of RGB gives out a perfectly neutral grey. This is the case, may I say it again, of the L*a*b* color space but also of sRGB, Adobe RGB 98, DonRGB, ECI-RGB, Prophoto ... The sRGB color space, for example, was created as the lowest common denominator in the world for computing and digital imaging. ALL MONITORS AND PRINTERS IN THE WORLD KNOW HOW TO REPRODUCE IT. As it does not contain all the L*a*b* colors but only about 90% of the ones present in our photographs, all recent devices know how to deal with it. We will see in the chapter on image conversion why it is important that the RGB color spaces be significantly smaller than the L*a*b* color space and why, therefore, we created smaller colorimetric space than the L*a*b* color space and yet of different sizes.
A color space is a broader or smaller set of colors, with more or less defects. The biggest of all is the La*b* color space (colors a human being is able to see) and the best known is sRGB, the smaller common denominator to all devices on the market.
There are two main categories of color spaces:
The ones relative to devices: it means the set of RGB colors that a camera, a monitor or a printer is able to interpret, display or print. It is smaller than the L*a*b* space and most importantly, it contains all the defects and technical limits of the device. There are as many devices color spaces as there are devices. They're also called ICC profiles.
And the ones that are not relative to devices: they are virtual, more or less large and they are perfect. They do not depend on one device and are therefore perfect colorimetrically speaking. There are thus much less than ICC profiles. A good color space is not necessarily a very large color space. The most well-known are: sRGB, Adobe RGB 98 and Prophoto. Independent color spaces are mainly used as color workspaces in Photoshop and especially in your images.
The right color space is the color space you need! If you're shooting or working with very saturated colors, using the HDR technique for instance, I need wide gamut spaces as they're called, broader than sRGB, like ProPhoto. If you like "normal" colors though, not especially saturated, sRGB will be just fine! Trust me, try it and you'll see!
As we have talked about them on this page, it is time to go further on ICC profiles
This site is entirely free of charge and yet almost free of commercial ads. Why? Well because they're ugly! So apart from this banner, you won't find any other commercial spaces. But you have to admit that advertising is pretty profitable and that it enables many websites to keep on living.
So I asked myself the question: what can I do since this website can only continue its growth generating regular incomes, like all other websites. How can I keep on giving access to all this information for free, which I've historically been doing since April 2002 (in french)?
Well thanks to affiliate programs. Meaning that every time you click links from this site towards my partners', Amazon.com, B&H, etc, they know you're referred by me and give me a small percentage on the sale. It doesn't cost you anything! Everyone wins: you, them, me. And also, feel free to contribute here: