After calibration, test patterns from the Video Essentials and Avia test DVDs indicated a slight green shift at the darkest end of the 50H81's gray scale. Black-level restoration was about average, with the black level rising slightly as the average picture brightness increased. Some internal reflections visible onscreen (on black or near-black images) were no worse than average for a rear-projection set. Overall, the blacks were good. As is typical for CRTs, the blacks were so deep that measuring the contrast ratio using our professional Minolta light meter proved impossible; the meter simply would not respond low enough to provide a reading for the minimum black level.

As noted in the review, the Movie mode appeared to disable scan-velocity modulation and was used for all of my viewing. The phosphors in the set's CRTs produced color very close to the HDTV standard: blue was off by a maximum of about 8%, but red was within 3%, green within 1.3%. The color decoder appeared to be reasonably accurate: red push, though present, was moderate and did not significantly affect the color quality. Overscan measured about 5% at the left and right and 3% at the top and bottom.

I noted only three significant limitations with test patterns. First, there appeared to be a distinct falloff in the frequency response above about 5mHz, as judged subjectively by viewing the luminance sweep (chapter 17-23) on Video Essentials. But there was also a series of brighter vertical bands on this pattern, beginning at about 3.2mHz and recurring about every 0.3mHz, that indicated peaks in the response. This characteristic was present with progressive (480p) and interlaced (480i) sources, but not with a 1080i input. The Avia resolution test showed no detail in the response at 6.75mHz (which represents the full 540-line resolution of the DVD format). The horizontal resolution pattern on Video Essentials, in fact, showed no significant detail above about 450 lines. This resolution was virtually the same as what I'd observed on Toshiba's older TW40X81, and reconfirmed on that set for this review. The main difference is that the TW40X81's response falls off smoothly, without the response peaks above 3.3mHz observed in the 50H81.

Second, while 3:2 pulldown recognition is listed as a feature of the 50H81's onboard scaler, it did not function well. In chapter 20-2 of the Snell & Wilcox Zone Plate Test on Video Essentials, the pattern lock that indicates functional 3:2 pulldown never fully engaged before the bouncing ball changed direction.

And third, though the lines on crosshatch patterns were straight and true, the sizing of the image was a little off with hi-def images. That is, the picture was vertically stretched just a bit. This was very subtle, and I suspect most viewers would not notice it. As far as I could determine, there is no way to adjust this out without causing pictures from standard-definition sources (which otherwise looked fine) to distort a little in the opposite direction. There is a Width control in the service menu, but it controls HDTV and NTSC sources equally. On the other hand, a problem like this, since it depends on production-line setup, could be specific to our sample and not a generic flaw.

While the rolled-off high-end response, response peaks, and 3:2 pulldown issues were never fully resolved, I finally arrived at a satisfactory compromise on the geometry issue by adjusting the Width control in the service menu to midway between the optimum settings for 1080i and 480p/480i. Neither setup was then perfect, but the deviations for both were below my nuisance threshold.

The Before and After gray scales (Warm position; the others produced significantly higher color temperatures) are shown in the accompanying graph.—TJN