THE SAME, BUT DIFFERENT
Mar 1, 2001 12:00 PM,
Peter Putman, CTS
With the long-term shift from picture tubes to flat-screen monitors, the issue of color calibration takes on new importance.
FUNNY THING ABOUT COLORS: NO TWO people see them the same way. The same can be said for professional color-analysis tools when it comes to calibrating flat-matrix electronic displays. LCD and DLP projectors, plasma display panels and LCD monitors present a real calibration challenge; they use sources of illumination quite different from the sun, filament lamps and cathode-ray tubes.
Over the past year, informal debates and discussions on this very topic have appeared in print and on trade and consumer Web sites. People who perform color analysis and calibration of monitors and projectors for a living now suspect that their expensive instruments can be fooled by short-arc projection lamps and fluorescing displays.
With the long-term shift from picture tubes to flat-screen monitors, the issue of color calibration takes on new importance. This is particularly so where dealers and systems integrators are installing plasma display panels to replace CRT monitors, or front LCD and DLP projectors in homes and businesses.
Initially, I was a little skeptical of claims that conventional color analyzers weren’t reliable for such calibrations. After all, color is color! Specific mixtures of red, green and blue will produce color shades that can be assigned specific x,y coordinates on the 1931 and 1976 CIE color charts. (Luminance values add a z component, or intensity, to these 3-D coordinates.)
While the “color is color” argument holds true technically, the problem arises in the type of color sensors that are used to measure specific values of red, green and blue. The most accurate device for color measurement is a spectroradiometer. This instrument determines precise x and y coordinates for any additive color mixes, so it can differentiate between the red you’d see on a CRT display and the red generated by a metal-halide lamp.
Spectroradiometers can also determine levels of red, green and blue, which is important to know when adjusting these levels for specific white balance measurements such as daylight (5,400°K) or NTSC white (6,500°K). As you might expect, such devices are not cheap — they range in price from just under $10,000 to over $20,000.
That’s not a practical option for someone performing ISF calibrations or setting up banks of plasma displays for a trade show. A smaller, hand-held meter is more appropriate: one that uses tri-color sensors and can provide not only RGB values and color temperature, but can also show you which direction you need to move in the 3-D CIE color “space” to achieve the desired white balance.
Many new meters have come to the market for CRT calibration in the past few years. All of them can be used to measure and calibrate flat-matrix displays, provided some sort of correction is dialed into the meter. This correction could be a look-up table for specific models of plasma, LCD monitors or LCD/DLP projectors.
Cliff Plavin, owner of Progressive Labs in New Jersey, has developed a color analyzer that uses a notebook computer to do the number-crunching. (The tri-stimulus color-sensor head plugs directly into the laptop’s serial port.) Conversations with him prior to and at this year’s CEDIA Expo were part of the reason I decided to conduct tests of several color analyzers on plasma displays.
Further conversations with John McCasland of Minolta Corporation revealed a need for handheld, incident-light color temperature meters that could dependably read RGB values from LCD and DLP front projectors, as the large-venue projection market is dominated by these imaging technologies.
As a result, I set up a small test lab to conduct side-by-side color readings of six flat-matrix displays: A Princeton AF3.0HD CRT monitor, a Pioneer PDP-M502X 50-inch plasma panel (fluorescing light), Sony’s VPL-VW10HT widescreen LCD front projector (short-arc UHP lamp), an NEC MultiSync MT1055 desktop LCD projector (short-arc UHP lamp), an NEC XT5000 3-chip DLP projector (xenon short-arc lamp) and a Mitsubishi 18-inch 1280 ? 1024 LCD monitor (fluorescing light).
Keep in mind that this test was not conducted to reveal any shortcomings of the color analyzers I used. Rather, the point was to show how different instruments “see” color from a variety of emissive and transmissive light sources.
For these tests, I designated a Minolta CS-1000 Spectroradiometer as my reference color analyzer. Minolta also supplied a handheld CL-1 color temperature/luminance meter. I took additional readings with FSR’s Color Analyzer, a notebook-PC-based color measurement device; and employed two Sencore color analyzers — the Color Pro 288 handheld meter and the Color Pro 290 color analyzer, also a notebook computer display.
All of the meters were used to measure the white balance and individual red, green and blue settings on a Princeton AF3.0HD 32-inch CRT monitor. I’d expect to see the smallest variations in readings with this type of display, as all of the analyzers and meters can read SMPTE C phosphors.
The only meter that had substantial errors in measurement was the CL-1, which wasn’t set up for a contact measurement. The CS-1000 read color temperature higher than all other analyzers (7,607°K) by about 1,000°. Among the color analyzers, the FSR and two Sencore units were very close — within a range of 900°K.
Excluding the CL-1, individual color coordinates were close across the board. The largest variations were seen in red and blue measurements, indicating possible low sensitivity to these colors in the sensor heads. Even so, it’s safe to say that all of the color analyzers could be set up to within a few hundred degrees of each other and the CS-1000 (again, excluding the CL-1).
Plasma panels are unique beasts in that they function like expensive fluorescent lights. Supposedly, the ultraviolet burst in each pixel renders a color shade differently than an electron beam would, so I expected to see noticeable differences in CIE coordinate readings between the meters. (All readings were taken with a 60Hz image-refresh rate.)
Surprise! None of the readings varied by more than 1,100°. Throw out the readings from the CL-1 (again, set up for a non-contact incident reading) and the range drops to 848°. The CS-1000 measurements were done in manual refresh mode, using 1/60 as a refresh rate. The FSR and Sencore CP288 were very close in readings, but still about 800° high from the CS-1000.
UHP FRONT PROJECTOR #1
For this test, I used the Sony VPL-VW10HT. UHP and other short-arc lamps used in LCD and DLP projectors are strong in blue-green spectral output, and weak in red, orange and yellow. I would assume that unless the CL1, FSR and Sencore analyzers were pre-calibrated for the spectral curve of this kind of lamp, they would have gross measurement errors.
That’s pretty much how things turned out. The CS-1000 indicated a color temperature of 6,790°, only 240° higher than what the incident-light CL1 meter showed. The FSR unit was quite a bit higher at 9,840° with the Sencore CP288 higher still at 10,011°. The CP290 couldn’t resolve the color temperature at all. Blue coordinates didn’t vary significantly from meter to meter, but red and green coordinates were all over the place.
UHP FRONT PROJECTOR #2
This time, I used the NEC MT1055. The NSH lamp used in this projector burns even colder than a UHP lamp, resulting in a picture with less green and more blue. I figured this lamp would give all the meters fits. Red and yellow spectral output is reduced even more than a UHP lamp, so it’s hard to warm up the image using internal RGB drive settings.
The CS-1000 showed NEC’s “out of the box” white-balance adjustment to be slightly over 8,000°K, very close to the CL-1. Normally, you’d use a special neutral density filter and correction factors with such a bright light source, but the filter shipped by Minolta was for a different lens. So, I used generic neutral density material, Roscolene.
The FSR readings from this projector ran nearly 4,000° higher, followed by the CP288 (over 5,000° higher). Once again, the CP290 couldn’t resolve the color temperature. In this group of measurements, green and blue coordinate errors were the culprit, and red measurements didn’t vary as much.
XENON FRONT PROJECTOR
The NEC XT5000 projector is too bright for the CS-1000 to measure without the correct neutral density filter, so I left it idle. The Xenon “bubble” lamp used in this projector has much more red, orange and yellow spectral output than do metal-halide and UHP lamps.
The variations across the board amounted to no more than 460° — hardly worth worrying about. More variation was seen in the red and green channels than in the blue channel. Personally, I’d use a handheld incident light meter before a contact-type color analyzer to calibrate a front projector, because the handheld meters can handle these high brightness levels.
LCD DISPLAY MONITOR
LCD monitors can be a bear to set up correctly, as they use a “cold” light source similar to a fluorescent lamp. I expected all of the meters (except the CS-1000) to have trouble reading the Mitsubishi LSA 820W accurately.
Surprisingly, the variation in readings totalled only 722°. All three contact-type analyzers read the actual color temperature too low, not factoring enough blue into the equation. Variances were seen in both red and green channels as well.
You can use several of these meters out of the box for adjusting almost any flat-matrix display (except those using UHP and mercury short-arc lamps) and get pretty close. Most people can’t see color shifts between 500° and 1,000°, and it takes a really trained eye to sense swings of less than 500°.
Still, it makes sense to have look-up tables for all flat-matrix sources, if only to cut that margin of error by 50%. If you are calibrating a plasma display for home theater use, you’ll need a color analyzer that reads (and displays) in three axes (x, y and z), so you can tell which direction you need to move with R, G and B adjustments. The FSR and Sencore CP288 will let you do that. Look-up tables for different plasma panels are being compiled now for the FSR analyzer, and I would suspect Sencore is also planning to compile similar data for plasma and TFT LCD monitors.
The CP290 is better suited for calibration of CRT projectors and monitors. This meter works best with equal-energy color sources, as does the CL-1, which can measure thousands of lux. New hand-held meters coming to the market (like Minolta’s CL-2) will have look-up tables for short-arc lamps to permit closer and more accurate color calibration of front projectors.
Peter Putman owns PHP Communications, Doylestown, Pennsylvania. He is the author of The Toastmasters Guide to Audiovisual Presentations and reviews large-screen displays and computer/video interfaces.