Doug Kerr
Well-known member
When we speak of the gamut of a color space (color gamut is more completely descriptive), we mean the "range" of colors that it can represent. Similarly, when we speak of the (color) gamut of a display system, we mean the "range" of colors it can produce.
Color is a three-dimensional property (in the mathematical sense - it takes three numbers to describe a color). There are of course various systems of coordinates in which we can describe a color (and any given color space uses a particular one). To "plot" a color in some chosen three-dimensional coordinate system, we must really work in three-dimensional space - that is, we can;t really do it on paper, but require some kind of "solid model" capability.
The "gamut" of a color space or display is thus a three-dimensional object, bounded by a three-dimensional surface.
Of course, just as we can "by projection" make a two-dimensional representation of a three-dimensional thing (we do it every time we take a photograph!), we can make a two dimensional representation of a three-dimensional "gamut solid", and we often see these. If these are cleverly shaded, they give us a good ideal of the shape of the "gamut solid", but that doesn't necessarily make it easy to figure out what that means to us (although color wonks get use to figuring that out).
But we often see what are said to be graphs of the gamut of a color space or a display that are in fact just two-dimensional. They are often plotted on the familiar CIE x-y chromaticity plane (the one on which we often see the "horseshoe-shaped" human chromaticity gamut). Sometimes we are led to believe that these actually only describe the color gamut of a color space or a display system.
But they don't - they only show a chromaticity gamut - and I emphasize "a", since such a color space or a display device does not normally have a universal chromaticity gamut.
To jump to the punch line, we must note that for a typical display, for example, it cannot generate colors of all the chromaticities enclosed within the boundary of the "gamut" plot we usually see over the full range of available luminance. (I am not referring to over the range of the "brightness" setting of the monitor - I an referring to the relative luminance of individual colors we may wish it to produce.)
A little thought will reveal why this is.
Imagine an RGB display system When each of the three "primary channels" is "wide open", the display usually produces light of the display's native refernece white chromaticity at a certain luminance.
Suppose we want instead another color of this same luminance but with a different chromaticity - one toward the "red", for example. Well, we can't do that. To do so, we would have to increase the output of the "red" channel while decreasing the outputs of the "green" and "blue" channels to keep the luminance constant. But we said the red channel was already "wide open".
Thus, to have the display generate a color with any other chromaticity than its refernce white, we must be willing to accept a luminance less than the one we can get for the reference white chromaticity.
So what is the significance of the single chromaticity gamut plot that we often see? Well, it shows the boundary of the range of chromaticities the display can generate (or the color space can represent) at some particular luminance (or, in many cases, for any luminance not above a certain point).
For example for the sRGB color space, the familiar triangular plot of its "gamut" is the boundary enclosing the chromaticities of the colors it can generate whose luminance is not over 7.2% of the luminance of the "full reference white" color!
Well, that's pretty limiting. What if we are interested in higher luminance colors than that?
Well, here we see some chromaticity plots on the familiar CIE x-y chromaticity diagram:
The red triangle is what we often see as the "gamut of the sRGB color space." As we see from the legend, all the chromaticities enclosed by that triangle are available so long as we aren't intersted in colors whose relative luminance is over 0.072.
What if, for example, we are interested in colors whose relative luminance is 0.500 (50% of maximum). Well then we can have any chromaticity we want so long is it is inside the blue line!
Now suppose we want some colors whose relative luminance is 1.000? Well we can have one color: the one with the reference white chromaticity (shown by the black dot).
Of course, there is a similar situation for the gamut of a display device.
So lets be careful that, shown a nice triangular pattern on the CIE chromaticity diagram for some display system, we don't say, "Wow! That's a really large portion of the human chromaticity gamut!".
Color is a three-dimensional property (in the mathematical sense - it takes three numbers to describe a color). There are of course various systems of coordinates in which we can describe a color (and any given color space uses a particular one). To "plot" a color in some chosen three-dimensional coordinate system, we must really work in three-dimensional space - that is, we can;t really do it on paper, but require some kind of "solid model" capability.
The "gamut" of a color space or display is thus a three-dimensional object, bounded by a three-dimensional surface.
Of course, just as we can "by projection" make a two-dimensional representation of a three-dimensional thing (we do it every time we take a photograph!), we can make a two dimensional representation of a three-dimensional "gamut solid", and we often see these. If these are cleverly shaded, they give us a good ideal of the shape of the "gamut solid", but that doesn't necessarily make it easy to figure out what that means to us (although color wonks get use to figuring that out).
But we often see what are said to be graphs of the gamut of a color space or a display that are in fact just two-dimensional. They are often plotted on the familiar CIE x-y chromaticity plane (the one on which we often see the "horseshoe-shaped" human chromaticity gamut). Sometimes we are led to believe that these actually only describe the color gamut of a color space or a display system.
But they don't - they only show a chromaticity gamut - and I emphasize "a", since such a color space or a display device does not normally have a universal chromaticity gamut.
To jump to the punch line, we must note that for a typical display, for example, it cannot generate colors of all the chromaticities enclosed within the boundary of the "gamut" plot we usually see over the full range of available luminance. (I am not referring to over the range of the "brightness" setting of the monitor - I an referring to the relative luminance of individual colors we may wish it to produce.)
A little thought will reveal why this is.
Imagine an RGB display system When each of the three "primary channels" is "wide open", the display usually produces light of the display's native refernece white chromaticity at a certain luminance.
Suppose we want instead another color of this same luminance but with a different chromaticity - one toward the "red", for example. Well, we can't do that. To do so, we would have to increase the output of the "red" channel while decreasing the outputs of the "green" and "blue" channels to keep the luminance constant. But we said the red channel was already "wide open".
Thus, to have the display generate a color with any other chromaticity than its refernce white, we must be willing to accept a luminance less than the one we can get for the reference white chromaticity.
So what is the significance of the single chromaticity gamut plot that we often see? Well, it shows the boundary of the range of chromaticities the display can generate (or the color space can represent) at some particular luminance (or, in many cases, for any luminance not above a certain point).
For example for the sRGB color space, the familiar triangular plot of its "gamut" is the boundary enclosing the chromaticities of the colors it can generate whose luminance is not over 7.2% of the luminance of the "full reference white" color!
Well, that's pretty limiting. What if we are interested in higher luminance colors than that?
Well, here we see some chromaticity plots on the familiar CIE x-y chromaticity diagram:
The red triangle is what we often see as the "gamut of the sRGB color space." As we see from the legend, all the chromaticities enclosed by that triangle are available so long as we aren't intersted in colors whose relative luminance is over 0.072.
What if, for example, we are interested in colors whose relative luminance is 0.500 (50% of maximum). Well then we can have any chromaticity we want so long is it is inside the blue line!
Now suppose we want some colors whose relative luminance is 1.000? Well we can have one color: the one with the reference white chromaticity (shown by the black dot).
Of course, there is a similar situation for the gamut of a display device.
So lets be careful that, shown a nice triangular pattern on the CIE chromaticity diagram for some display system, we don't say, "Wow! That's a really large portion of the human chromaticity gamut!".