Running The Gamut(s)
Color measurement for electronic displays is getting easier and easier.
ONE OF my favorite bugaboos regarding electronic displays is the quality of the color they can produce. Not all display technologies are up to the task — phosphor-based (CRT, plasma) displays and projection engines with dichroic filters generally do a better job than LCD monitors with cold-cathode (fluorescent) backlights.
But how much better? What constitutes “white” on a plasma monitor, versus “white” on a DLP projector? How well does your new monitor track a grayscale? Is it really that important to set the white balance on a projector?
These are all good questions, and they would be difficult, if not impossible, to answer without the help of color analyzers. These products are becoming more powerful and affordable with each passing year.
I recently had a project in-house that called for quantitative performance measurements from three flat-panel monitors (two plasmas and one LCD). My six-year-old color analyzer system — while it did yeoman duty for me several years ago — just wasn't powerful or accurate enough for this particular job. After some trade show and online research, I decided to purchase Datacolor's ColorFacts 6.0 bundle. This particular package, formerly sold by a company known as Milori (since bought out by Datacolor), consists of a tri-stimulus color sensor head, USB connecting cable, and a software package with lots of macros and wizards. It supports fast enough refresh rates to keep up with plasma displays and can be used in contact and non-contact modes. Because I already own a Minolta CL-200 chroma meter for incident (illuminance) readings from projectors, the ColorFacts system would be put to work exclusively for contact (onscreen) measurements of plasma, LCD, and rear-projection displays. One problem with older color analyzer designs is their inability to accurately measure color from emissive, transmissive, and even pulsed microdisplay technologies, which made my older analyzer difficult to use consistently on anything other than a CRT display. A separate reading was often required from a precision spectroradiometer to determine absolute values of red, green, and blue for plasma, LCD, UHP, xenon, and other light sources. That data became a reference lookup table to use as a baseline.
The ColorFacts system supposedly eliminates that need as it's at home with a wide variety of display technologies. That was borne out when I generated a color primaries chart using my reference Princeton AF3.0HD CRT monitor, then reset the instruments and took readings of the one LCD and two plasma monitors, comparing their color gamuts to the Princeton's reference gamut. ColorFacts was accurate enough to show that the test monitors came up short with one or more primary colors, although I had a close match to the value of green in the REC709 color gamut with one of the plasmas. I could clearly see the effects of different phosphor and color filter formulations as I plotted CIE “tongue” diagrams for each display. The value of this information was apparent when I later tried to calibrate each monitor to a specific color temperature (D6500) for showing some HDTV 720p and 1080i clips. Not all manufacturers provide access to the required RGB contrast (and sometimes brightness) adjustments.
With one plasma monitor, a special service remote control had to be sent from the factory so I could make the tweaks. On the other plasma display, it was a matter of getting the service access code to tune it up, while the LCD monitor provided RGB gain settings in its USER mode.
Color primaries are just the start. What's more important is how a display tracks color temperature over a wide grayscale. For this test, I used a series of gray windows from 20 IRE to 100 IRE, and plotted the x,y coordinates and resulting color temperature at each step.
Figure 1 shows how each display did with its factory calibration, and Figure 2 shows how they measured after I performed some tune-ups. Quite a difference! But it would have been impossible to correct these white balance plots without an accurate color analyzer. Some people can get close with a reference gray card, but I don't trust my eyes that much.
Degrees K vs. IRE Steps. Figure 1. Pre-calibration grayscale track. (orange=plasma 1, red=plasma 2, blue=LCD, yellow=D6500 Reference)
Degrees K vs. IRE Steps. Figure 2. Post calibration grayscale track. (orange=plasma 1, red=plasma 2, blue=LCD, yellow=D6500 Reference)
ColorFacts 6.0 also has a few “wizards” that can compile a wide range of display performance data, useful for plotting things such as Gamma. One of my plasma displays had its factory Gamma setting at 2.2, but subsequent brightness measurements with the stepped gray windows revealed the actual Gamma was closer to 1.0.
I could also plot the linearity of each color channel as it came out of black and proceeded up the grayscale to 100 percent white. Theoretically, the levels of each color should remain constant, but in practice you might find that at certain grayscale levels, one color is more or less saturated than others. That can make it a chore to achieve consistent white balance at low gray levels.