04/05/2006
On Monday I heard that later this week Mitsubishi will be showcasing some rear-projection TVs based on Texas Instruments' digital micromirror (DMD) DLP technology. That in itself would not ordinarily elevate my heart rate above 60, but mine has moved into the anaerobic zone since I also heard that the light sources for the new screens will not be the usual incandescent projector bulbs but lasers! Aside from having all the cachet of being far more up-to-date and high-tech than the usual light bulbs — invented way back in 1879 by Edison two years after he had invented the phonograph and about a decade before he was to make important contributions to early movie technology — the use of lasers in a video projector can have several important benefits. Although I won't be attending Mitsubishi's press event, I'll speculate here on what the company could do with the technology under ideal conditions.

If Mitsubishi has solved the traditional problem of solid-state lasers — limited lifetime when driven hard as they would have to be in a video projector — they might drastically reduce projector power consumption. After all, a light bulb's peak output is in the infrared, which is just heat energy, invisible and unnecessary for video reproduction. And because of the purity of laser colors, you ideally would not have to filter any of their output with a color wheel — a light-wasting proposition — to get the red, green, and blue primaries necessary for a TV picture. Presumably Mitsubishi would simply turn each laser on when its color was needed and off when it wasn't. So lasers could greatly reduce a DLP projector's energy consumption and the bulk of its light engine by diminishing heat-sink and fan requirements and eliminating the color wheel.

When it comes to the primary colors themselves, lasers operate in that blessed realm at the outer edge of the classic CIE color-space diagram's blobby triangle representing all the colors we can see. Note that this diagram (and others like it) represents but does not reproduce all visible colors, since that would be impossible on any computer or video monitor, much less a printed page. That's because the primary colors typically produced by video and computer monitors stand approximately at the three vertices of the straight-sided inner triangle shown here. All the colors it is possible to reproduce with those three primaries lie on or within the inner triangle, which does not include some areas of the total color space. In particular, the highly saturated colors near the edge of the color space, especially in the deep greens and blue-greens, aren't reproducible with standard video primaries. Furthermore, the inner triangle represents ideal video performance; the actual primary colors generated by any particular display may deviate substantially from the video standards. The reason has usually been to increase screen brightness, but nonstandard primaries are also likely to constrict the reproducible color gamut and to distort the colors that are displayed.

If the laser wavelengths are chosen properly — and there may be technical restrictions on what wavelengths Mitsubishi is able to use — the color gamut can increase substantially. If, for sake of argument, Mitsubishi has green at 520 nanometers (nm), blue at 470 nm, and red at 620 nm, the range of laser-reproducible colors (the triangle between those three points) will include the entire video color space and then some. Let's hope that Mitsubishi has managed to find lasers close to these wavelengths.

If it has, some interesting questions arise. Accurate reproduction of video signals requires sticking to the standard video color space and displaying its colors exactly, even though that space itself may have warped the colors it encodes. For example, normal video equipment will probably record an intense blue-green seascape such as you see around tropical islands as a less saturated blue-green, since the natural color lies outside the video color space. With an expanded color space available, designers may be tempted to somehow warp the colors back to where they were in the original scene, even though a video signal contains no information that would allow such a transformation. The opportunities for creative color distortion are enormous, and I just hope that if Mitsubishi does give in to them in its laser-driven DLP projectors it also supplies a back-to-normal-color button. As it is, if the company does everything right, "normal color" could come out dead-on true to the original signal.

Lasers' fast switching times lend themselves to playing some tricks in pursuit of higher contrast ratio, seemingly the fetish-spec of the hour. The light bulbs in today's DLP sets are always on and going full blast (hence the heat, and the noisy fans in some models), and even when reproducing an all-black signal there is usually enough stray light bouncing around a DLP optical system that the projected image isn't a pure black. Since lasers respond much faster than light bulbs, it is theoretically possible to turn down their intensity on a frame-by-frame basis when reproducing dark images to decrease the amount of stray light, possibly down to invisible levels, thereby increasing contrast ratio. They could even be turned off entirely when an all-black frame is shown to obtain a ridiculously high "sequential" contrast ratio (a spec derived by measuring light output from all-white and all-black frames displayed in sequence). But the problem has always been to reproduce deep blacks when the image isn't entirely dark, as in a scene with very bright highlights and very deep shadows, where laser-intensity modulation won't work. I'm eager to see how Mitsubishi has attacked this problem, too.

See the previous David's Dartboard entry.

Back to Homepage
What's New on S&V