TEST BENCH FOR WEB David Katzmaier

Color temperature (Low color temperature and Natural mode before/after calibration) Low window (20-IRE): 5,779/6,457 K High window (80-IRE): 5,848/6,439 K Brightness (100-IRE window before/after calibration): 136/43.1 ftL

The Natural picture preset on the WD-52627 automatically engages the Low color-temperature mode, which measured consistently lower (redder) - an average of about 670 K lower than the standard of 6,500 Kelvin. The grayscale remained extremely consistent after calibration at all light levels, varying by only 45 K from the ideal. (Calibration needs to be performed by a qualified technician, so discuss it with your dealer before purchase, or call the Imaging Science Foundation at 561-997-9073.)

Like most DLP televisions, the WD-52627 was extremely bright out of the box, and even the Natural setting comes with contrast set at maximum. I tamed it for critical viewing in the dark, which made the image bearable. The set reproduced the full range of black to white via all of its inputs, including HDMI. Color decoding out of the box was good, with a 5% bias in red and a -5% bias in green according to the color-decoder check from the Sencore VP403 HDTV signal generator, but after adjustment it was close to dead-on. Brightness uniformity was good, with minor hot-spotting toward the middle of the screen.

Convergence was excellent and overscan negligible at an average of 2%, although my review sample did exhibit some anomalies in geometry. The upper-left corner of the screen bowed outward slightly - an effect visible primarily on test patterns and something Mitsubishi warned was a symptom of our early sample. I also saw very slight pincushioning along the right and left edges of images, which was most noticeable when watching 4:3 material.

Black-level retention was relatively good; the level of black remained consistent in all but the most demanding test patterns from the Avia DVD. Even with sharpness properly adjusted, some edge enhancement was visible, which showed up as faint halos around onscreen objects, but this was difficult to spot without test patterns. Sharpness also had a major effect on resolution and high-frequency video noise. I settled on a sharpness setting of 3, with noise reduction set to standard.

The set was unable to accept a 720p signal via its HDMI input, although the component-video input did work with 720p. This was said to be another abberation of our early sample; Mitsubishi indicated the set was spec'd to accept 1080i or 720p (though not 1080p) on the digital inputs and that full production samples should behave properly. Multiburst resolution patterns from the Sencore and the 1080i version of Digital Video Essentials via all inputs revealed that the set could not resolve every detail of 1080i, but the patterns were noticeably more detailed than those on the comparison 720p set. HDMI was noticeably sharper than component video and about the same as FireWire.