Is anyone else using the Panel alignment feature to correct for panel alignment errors (convergence)?
When I first turned my unit on there were obvious panel alignment errors, immediately visible on the internal menu text and graphics. I reckon I had a green fringe on the text menus of at least one pixels width. I went into the Panel alignment function and selected "Shift" which works across the entire image as opposed to "Zone" which applies corrections to just a small part of the image. I figured "Shift" would introduce less artifacts. My unit needed Red to be shifted 5 to 6 clicks in each axis. Blue needed a couple clicks in one axis. The result was perfectly acceptable apparent convergence on the test grid, with almost no perceptible color bleed even up close. I did not see any obvious artifacts when viewing regular material.
However, on some of the 1080P test patterns on the AVS Set-up disc, I saw something very odd:
When I display a pattern with alternating black and white lines, of 1 pixel widths, the white lines show up as perfectly even RED lines! This must be a product of the panel alignment adjustments. While, I do not see this effect with regular material, I wonder if I am losing perceived sharpness.
I was able to verify that it is the use of the panel alignment feature that is causing the white lines to turn red on the single pixel-wide lines on the test pattern. When I turn panel alignment off, the thin (1-pixel) wide white lines turn from red back to white. However, I can then see the color fringing on the Sony menu and on wider white lines. The color fringing is very noticeable close to the screen, is still just about visible from 10 ft away and disappears at 16 ft back from the screen. As things stand, I can leave panel alignment off and have some visible color fringing on certain scenes (and the menu) or turn it on and run the risk of the type of artifacts that i experienced with the one-pixel width test pattern lines.
The spec sheet for the projector mentions that the panel alignment acts in 1/10 th of a pixel increments. I did verify that if I set the offset to exactly 10 clicks (I currently use 5 or 6 clicks) the artifacts on the pixel-wide zebra lines disappears, but then the color fringing returns, although the green and red fringe have swapped positions. Also each click seemed like it was much more than 1/10 of a pixel shift (maybe 1/4 of 1/5).
Is my unit the only one with panel alignment error so large as to require using the panel alignment feature? If anyone else is using it, please report your settings and if you can test your unit with these test patterns, even better. Thanks.