Flat-Panel Color--Any Good?

Color LCDs are a mainstream technology for laptop computers, and some makers, including Compaq and Apple, have discussed using only color displays on their notebook computers. But are LCDs suited for careful, accurate color work? The short answer: They're usable, but just barely. LCDs will never be good color-proofing devices for WYSIWYG color printing, because of limitations of viewing angle, contrast, and size.

A CRT monitor propagates light over a wide angle--nearly 180 degrees. By contrast, if you look at an LCD from outside its 30-degree viewing angle, it's difficult to see anything on the screen at all. Essentially, if you rotate the screen around a vertical axis, at some point you'll see nothing. Up to t hat point, the image contrast will steadily decrease, which means that at each angle you view the display from, the color looks different.

Diminished contrast is also an issue because LCD contrast is so low to begin with, compared to CRTs. If power consumption isn't a concern, then LCDs can get fairly bright. Their low end, however, isn't black, because some backlighting seeps through the "off" pixels, reducing the color gamut of LCDs substantially. In addition, bright LCDs get hot, which makes colors drift.

Size is a killer. With windowing GUIs, people want to see more on-screen. A 14-inch LCD won't cut it; the market wants 15, 17, or 21 inches, and producing these is a problem of nightmare proportions. The typical VGA-grade, 9 1/2-inch active-matrix LCD panel has nearly a million transistors. CPU makers manufacture chips with several million transistors, using redundant designs and getting tens of small chips on a silicon wafer. However, LCD makers can't build in as much redundancy, and they c an process only a few panels on each glass substrate. To get a large display, they might be able to produce only one panel on a single substrate. With an LCD, each defect is a dead pixel. As size grows, the likelihood of clumps of dead pixels grows dramatically, and manufacturing costs and yields become unmanageable. In the spring of 1994, Sharp announced the world's first 21-inch active-matrix LCD. It produced a single prototype. Clearly, commercially feasible, large-screen LCDs are a long way off.

On the horizon, Texas Instruments is perfecting another flat-panel technology, DMD (Digital Micromirror Device), which puts millions of small mirrors on a chip. Here, too, it's hard to scale this technology up to large sizes, though it seems to have better potential than LCDs. However, there's another problem. According to Jack Roberts of Dataquest, "with LCD's, when pixels are dead, they're off. And this is less noticeable than if those pixels were always on. But with DMD, there's a 50 percent chance that a dead pixel will be off or on, so defects are more noticeable."