Hitachi 50VX915 Directors Series LCD RPTV Tests and Calibration
Out of the box, the "Black and White" color-temperature mode hovered just below 5500 kelvins—the recommended color temperature for black and white films. But it dropped down to 4800K at 20 IRE and peaked a little over 5500K at 40 IRE. The High mode was far too blue, measuring 11,000 to 12,000 kelvins. Standard was centered around 6000K, though it dipped to approximately 5600K at 20 IRE and rose to 6300K at 40 IRE. The Medium mode was also a bit erratic, but on balance it was the best of the lot and is plotted in the Before curve in the accompanying figure. The color temperature, even post-calibration (the After curve), was still uneven, with a sharp drop at the very lowest light levels and some peakiness in the brightness midrange.
This Hitachi was unquestionably the most difficult calibration I can recall performing. For starters, the service menu's white lettering litters the screen, blocks a fair portion of the calibration test patterns, and does not drop away when you're ready to take a measurement. You can exit the service menu every time you take a reading, but this is unacceptably tedious. Alternately, you can position or aim the measuring device a few inches off center, thus avoiding the white lettering. I chose the latter approach.
To make a very long story short, I made three attempts at calibrating this set. The red, green, and blue service-menu controls (separate adjustments are provided for both the top and bottom of the brightness range) were not only unpredictable in the amount of change they made with each step, but they were interactive as well. For example, a two-step change in the green-drive control would often make no difference at all in the reading, but an additional one-step move of the green control would then change both the x and y white point coordinates by as much 10%.
In the final analysis, roughly six hours of calibration efforts resulted in a gray scale curve (see After in the accompanying diagram) that was only slightly better than the out-of-box Medium setting (Before). I was unable to eliminate either the severe drop in color temperature below 30 IRE or the rise to around 7000K in the brightness midrange. A 10-step gray-scale test pattern, even post-calibration, looked very uneven, with a clear brownish shift at the low end, a subtle blue shift in the midrange, and even a subtle hint of yellow on some steps.
And the final calibrated result, at least to my eye, was not all that different from the Medium setting before calibration. In other words, the calibration did not produce a visible improvement that was worth the effort.
Of course, one cannot assume that the Medium setting on all 50VX915 samples will be the same as ours. The calibration results for MF's sample (not shown) were slightly improved over the out-of-box measurements. Unlike MF's set, my sample did not measure plus green in the Medium color-temperature setting before or after calibration.
My sample's red, green, and blue color points also departed from both the NTSC and HD standards (which are slightly different). The red color point was a little deficient of the required coordinates; its reddish-orange color was a bit too orange, a deviation that was visible with normal program material. Blue extended a little deeper into purplish blue than it should, but that was not evident to my eye. However, green was far outside the standard, adding a whole range of greens to the color gamut.
Isn't it good to have a wider range of colors in a video display? Color points that fall outside the required standard sound like a feature, not a bug. But accurate color depends not only on the correct color temperature—the color of all shades of gray from black to white—but the proper red, green, and blue color points as well. Color programming is produced to a standard (or should be), which calls for a specific mixture of established shades of red, green, and blue to produce a full spectrum of colors. If the color points of the playback display are different than those used to create the programming, all the reproduced colors will be inaccurate. They may look more vivid and exciting, and will stand out on the showroom floor, but they are still wrong.
The Hitachi's overscan (through an HDMI digital video input) was about 3% on all sides. I noted no serious color unevenness in different areas of the screen, even on a full white field—the most difficult test. The set's peak contrast measured 368 (34.3 fL peak white, 0.093 fL video black).—TJN