What's the difference between a spectrophotometer and a colorimeter? I've been told that a spectrophotometer is better, but I'm unclear on how and why.
You may have heard the Urban Legend that a spectrophotometer is reflective device (reads patches on paper), and a colorimeter is an emissive device (reads displays). Not actually true. There is another one, often combined with the first, that spectrophotometers are expensive devices, and colorimeters are cheap ones. Not necessarily true either; there are lab grade colorimeters that cost more than a small car, and spectros that cost less than $500. The the final myth is that spectros are more accurate than colorimeters. A laboratory grade spectro is usually more accurate than a consumer grade colorimeter; but then, a lab grade colorimeter would tend to be more accurate than a consumer grade spectro as well. So when you hear any of these myths, treat them accordingly.
The real difference is not emissive or reflective (both devices come in both types) or lab versus consumer grade (again, both device types come in both price ranges) but rather the method they use for reading color. A spectrophotometer breaks light up into a spectrum, using a color grating or similar system. Then an array of sensors reads each section of the spectrum, producing spectral data. This is ideal if you are analyzing the spectral emissions of a lightbulb, a star, or some other light source, which is why spectrophotometers are often used as scientific devices.
A colorimeter uses edge band filters, or some similar system, to separate light out into color components, and then fits those to matching curves based on the human eye, to produce color values in one or another three-value color space (XYZ, xyY, Lab, etc) based on what the human eye would see. This is ideal for matching the human visual response, but tells you nothing about data invisible to the human eye, such as emissive spikes at narrow points in the spectrum; thats spectral data, and requires a spectro.
The spectral data from a spectro can be recombined, through appropriate algorithms, to produce Lab, XYZ or other such values as well. Its more work, and there are some disadvantages to separating light into dozens and dozens of measurements, then adding them back together. So for luminance measurements, a spectro tends to be at something of a disadvantage. But for random color reading, it might (depending on the device, and the color) have an edge.
You might notice that none of the general descriptions above actually fit Datacolor's devices very well. Our display calibration devices (specifically the Spyder X) use several sensors with several different filters to obtain more information than a simple RGB colorimeter would obtain from the same lightsource, but far fewer bands than a grating spectro would obtain. This offers some advantages in terms of measuring unknown colors, without splitting the light into so many separate packets that its hard to generate accurate luminance data from it.
And Datacolor's print reading devices (specifically Spyder Print) uses a set of eighteen carefully selected LED lightsources, at six zones in the spectrum, to illuminate the print it is measuring, and provides, again, more data than an RGB colorimeter, but fewer data points than a typical spectrophotometer. This is why you sometimes see it referred to as a spectrocolorimeter, or even more precisely as an inverse spectrocolorimeter.
But the bottom line is how well the type of device matches the job at hand, and we find Spyder Print to produce measurements from measuring standard color targets that are virtually indistinguishable from what a consumer spectrophotometer produces when measuring the same target (less than one Delta E average, to get technical), while our screen colorimeters are generally accepted as the most consistent of the available display calibration colorimeters.
So you should feel free to choose the device you use for profiling based on its own merits, rather than worrying about which device type it may be.
C. David Tobie
Global Product Technology Manager
March, 29th 2010