Colorimetry

Colorimetry, or the measurement of color, is an important part of measuring a monitor’s performance. A lot of people assume that color accuracy is only important for photography or graphic design and art. That’s true, but the more accurate the color and wider range of colors that can be reproduced, the better the micro-contrast too. Now the color experts out there will tell me that local contrast and color accuracy are NOT the same thing and that's true, but ponder this: if an intense yellow color ends up being reproduced as a pale yellow, the overall image will also lose contrast.

Color accuracy is actually extremely important for gaming too. Think about how much time you spend making sure your GPU isn’t faking its anisotropic texture filtering. What if a game with washed-out and drab textures only looked like that because it was your monitor? When artists and level designers create textures and lighting, they’re making their artistic choices based upon what they see on their monitor. We surveyed a number of successful software developers ranging from junior start-ups with only one or two hits under their belt, to the studio powerhouses creating games that push the limits of NVIDIA and ATI’s flagship products and define the direction of the industry. Nearly all of them reported using color calibrated LCDs and CRTs as part of their design environment. They’re taking the time to make sure their games look their best… so should you.

Measuring color is easy – you can use a color meter... or a colorimeter. But which colorimeter? We used the X-Rite DTP94 colorimeter. The DTP94 is almost twice as accurate as the ColorVision Spyder used by some other review websites and is also better than the first generation Lacie Blue Eye colorimeters used by other reviewers. In fact, the DTP94 is the colorimeter of choice for Kodak’s Matchprint Virtual Proofing System and FujiFilm’s Level One Monitor Profiling Package.

Still, the DTP94 is just a piece of hardware. It is nothing more than a glorified ultra-low resolution digital camera that can accurately quantify the color being displayed. For example, you could measure how red the red looks and get a number back or measure the reddest red and greenest green that can be displayed. The catch is that since monitors are non-linear, we need to look at more than the reds, greens, and blues. We need to look at more than 3 colors, but which colors? Fortunately, that’s an easy question. The colors we’ll be testing are those made famous by the GretagMacbeth ColorChecker. This is a patch of 24 scientifically selected colored squares in a wide range of colors that represent natural objects such as human skin, foliage, and blue skies and have stood up to the test of time. Doing well with these 24 colors is predictive for doing well with all 16.7M colors.

The is the FiringSquad Reference Standard Color Target Image. This is the theoretical image that a perfect monitor should display

You do not need a calibrated monitor to take advantage of these special images. While it’s true that a perfectly calibrated monitor will display all of the reference colors correctly, the absolute color isn’t what’s important. You want to compare the differences between this reference image generated using Photoshop, and the images generated from measurements of the monitor we are evaluating. The relative differences between the monitor will tell you how the monitor looks.

Let’s pretend that the colors from “Monitor X” are too blue. On a calibrated monitor, the grays in the reference image will be perfectly neutral while the image for “Monitor X” will have a slight blue-tint. If your personal monitor is too blue to begin with then, the reference image might have a blue tint to it, but the image for Monitor X image will look that much more blue. If your personal monitor is too red to begin with, then the reference image might actually look red-tinged but it will still be clear that the Monitor X looks less red. Remember, think “relative difference” and not “absolute image.”

We allowed all of our monitors to warm up for one hour. We focused our attention on out-of-the-box accuracy for D65. There are two reasons for this: If you don’t own a colorimeter but still want accurate color, it might make more sense to spend a little extra on a monitor that is more accurate to begin with. Second, although you can measure and correct the color being displayed in 3D games and desktop applications, the color ICC profiles do not affect video overlay. You are stuck with the uncalibrated output when watching movies.