In Mitsubishi's 6-Primary color system additional segments for yellow, cyan, and magenta—usually referred to as secondary colors—are added to the color wheel. These new color wheel segments are driven by yellow, cyan, and magenta signals that are derived by the set from the red, green, and blue information in the source. In other words, in a conventional DLP the secondary colors are physically produced by a properly mixing the contributions of the red, green, and blue segments of the color wheel. In the 6-Primary system, the red, green, and blue are mixed electronically to produce yellow, cyan, and magenta signals, and these signals determine how the new, dedicated yellow, cyan, and magenta filters on the color wheel affect the projected image.

This technique is said to produce a richer color palette than can be derived via the usual mixture of red, green, and blue. But since the yellow, cyan, and magenta must still come from the red, green, and blue information contained in the source, and their precise coordinates in the color space are specified in the (similar but slightly different) NTSC and ATSC color standards, the accuracy of the overall colorimetry is ultimately determined by how correctly the set generates these colors.

Setup and Control
All the operating controls are located under a covered panel on the front of the set. They're also duplicated on the multifunction, programmable, but non-learning remote control. My only reservations about this device are the lack of backlighting and the absence of direct input access. The Input button calls up an on-screen Input Selection menu.

Setup of the WD-57731 should be straightforward for anyone who has dealt with a full-featured set before. It's also reasonably well explained in the manual for those who have not. A full set of clear, on-screen displays—Mitsubishi calls this the ViewPoint menu system—is provided. The set's ClearThought auto input sensing feature enables the set to automatically sense which inputs are connected, and displays icons on the menu only for those inputs (except ANT-1 and Cable-2, which are always displayed).

Once you've run through an automated search for the available channels (which won't be necessary if you rely on an outboard cable or satellite tuner), the Mits offers its own program guide, showing the channels saved by the set and current and future programs for each channel if that information is included in the broadcast signal. This is a nice feature, but isn't quite as useful as the program guides provided by most cable and satellite services.

The factory default settings for the video controls—Sharpness and Contrast at maximum, Noise Reduction at Medium (presumably to help compensate for the negative effects of a peak Sharpness setting!), Color Temperature on High, Picture Mode on Brilliant—produce a picture that only a dealer with a brightly lit showroom could love.

But that's easily fixed. Of the two Color Temperature settings, Low and High, Low was preferable by far. Later measurements confirmed that Low was closest to the D6500 standard, although calibration did produce a worthwhile improvement.

The Picture Mode control can be set to Brilliant, Natural, or Bright. I found little use for Brilliant or Bright; they would have been better named Showroom and Sun Porch! If I had used either for this evaluation, I never would have seen just what this set can do. Natural is the only way to go, but then I always do my serious viewing in subdued room lighting or, in the case of front projectors, in as dark a room as I can manage.

The Video Noise control offers four positions: Low, Medium, and High, plus Off. It works surprisingly well without unacceptable softening of the image. On more than one occasion it made the difference between me tolerating an annoyingly grainy standard definition picture on a poor cable feed and actually enjoying the program without trying to mentally tune out the video noise.

PerfectColor offers individual control over the intensity of red, green, blue, yellow, cyan, and magenta. A similar feature is currently popular on sets from many manufacturers. Adjusting any or all of these controls, even to their extreme positions, does not alter the color temperature in the slightest. But it does affect the intensity of the individual colors. I found no real benefit to this feature and left the controls at their default settings. Determining the color intensities in a video source is a creative decision that belongs at the production end, not in the final display. There are standards for the display; deviate from them and you're no longer seeing what the program creator intended.

The remaining controls—Contrast, Brightness, Color, Tint, and Sharpness, are self-explanatory. Unlike most sets, this Mitubishi does not present you with a full-screen menu that displays all these video controls together. Instead it cycles through the individual adjustments each time you push the Video control on the remote. While this isn't as flashy as a full-screen menu, it does keep most of the screen clear so you can see what you're doing when you tweak the adjustments.

One welcome feature is that the video adjustments may be made separately for each input. One quirk to watch out for: Whenever you disconnect an input and reconnect it, the auto input sensing feature assumes you want to start over. The saved video settings for that input are erased and the controls revert to the factory defaults. Also, for reasons I could never pin down, on several other occasions the controls spontaneously reverted to their factory settings. Fortunately, I had written down my preferred settings before this happened.

The High and Low color temperature settings may be calibrated separately in a code-accessible service menu, but apart from those two settings the color temperature is global across all the inputs.

The WD-57731 offers a wider range of aspect ratio selections than most sets. Mitsubishi calls them Display Formats, and their names can be a little confusing. For example, to display a correctly proportioned 4:3 image with black bars on either side, you don't select Standard—the obvious choice—you choose Narrow. To display a 4:3 letterboxed image that fills the full width of the screen, you select Expand, not Zoom. The Standard option is, however, the correct one for displaying a proper image from native 16:9 sources.

Two uneven stretch modes are included to reconfigure 4:3 programs to fill the 16:9 screen: Stretch and Stretch Plus. Both provide somewhat less distracting distortion (that funhouse look) than most other sets offering a similar feature. Stretch retains most of the image's vertical information, while Stretch Plus trades off a bit of the image at the top and bottom of the picture for less distortion in the middle.

Software and Warranty
A full page in the owner's manual is devoted to a license agreement for the software used in the set. Put succinctly, you are not authorized to access, analyze, disseminate, sell, lease, lend, lend lease, sublease, sublicense, fold, spindle, or mutilate the software used for any function within the set. A sign of the times.

Another sign: It takes a full page to describe the set's warranty—essentially 30 days on the screen and one year parts and labor on most everything else except some embedded software. It covers parts only and not labor on items designed to be replaced by the consumer, which likely includes the lamp. Interestingly, the warranty on the DMD itself allows for 0.01% defective pixels. That sounds peachy, but for those without their calculators handy, a 960x1080 DMD has just over 1,000,000 pixels. 0.01% of that is 100 pixels, which wouldn't be an acceptable flaw threshold even for the most non-critical viewer. But my experience suggests that a failure rate of even one pixel in a DLP display is extremely rare. It has been a remarkably reliable technology.

In Action
When I powered up the set a low-level whine was audible in my relatively small room, most likely from the rotating color wheel. But it was easily masked by almost any sound from the program source. The set also insisted, at first, on starting up with the antenna selected regardless of which input I was watching when I last turned it off. I disconnected all the inputs, reconnected them, and thereafter it came back on with the last input I was watching.

On my battery of specialty deinterlacing and scaling tests the set's internal video processing turned in only a fair performance. Subtle jaggies were visible on some of the tests, but not on others. DVD tests encoded without a 3/2 pulldown flag were slow to lock on and didn't always hold the lock, so you might find some improperly flagged DVDs that will not play well. A test for bad edit detection (an edit can interrupt the 3/2 pulldown sequence) showed some jagged edges. Noise was also more visible than usual in my noise tests, but the Noise Reduction control did an excellent job of cleaning it up without a serious impact on image detail.

Because of the above results and, just as important, some non-defeatable oversharpening and ringing visible on test patterns when the set was fed 480i, and, to a lesser extent, 480p (see "Tests and Calibration), I did much of my serious viewing with either native high-definition material or standard definition via an upconverting DVD player. This is not to suggest, however, that you absolutely must use the Mitsubishi with an outboard scaler and/or sources that do their own upconversion. Most users will not notice the problems, as they are far harder to spot on regular program material than on specialized tests and test patterns. But the set is capable of technically superior performance with 720p/1080i inputs than with 480i/p, whether the source is native HD or properly upconverted standard definition material.

In addition to the oversharpening visible with test patterns at 480i/p, there is also some oversharpening on 720p and 1080i as well. This is visible as edge haloes above and below the horizontal black lines on a sharpness test pattern—white lines that cannot be eliminated at any setting of the Sharpness control. The Sharpness control does, however, allow you to eliminate white lines adjacent to the vertical lines on a sharpness pattern. While this non-defeatable sharpening of horizontal lines was quite obvious on a test pattern, it was very hard to spot on normal program material.