Keystone wastes DLP pixels

[wernerj]

@lukehutch

What do you mean by this? DMD? Pixel shift mode? The DLP can definitely use all the available pixels, because it does so on the home screen with keystone correction disabled, but the outermost corners of the trapezoid with keystone correction enabled do not touch the outer edges of the DLP displayable area, they are set in from the side of the dark gray rectangle which is the DLP chip’s total display area. This is simply a waste.

The DMD (Digital Micromirror Device) is the part of a DLP-based light engine that contains the pixel-forming mirrors in the DLP system. Pixel shifting is how this particular DMD in this projector can display 1920x1080p video at 60Hz. The DMD itself contains a total of 1000x580 micromirrors where the center 960x540 can form active pixels. With a four-way actuator these form the 1920x1080 pixels you expect to see on the wall/screen, with an additional 40 pixels on each sides that always stay black. Your screen size when having keystone correction disabled seems to fill the entire active area, so I agree with you that something seems to be off with the correction. We already have learnt about the lack of linearity correction for 4-corner correction, that’s being worked on.

@lukehutch

I found that my projector’s horizontal alignment calibration was off by more than 45 degrees. I found the option in the settings to re-calibrate it. But if you want to see what I mean, hold your projector pointing 45 degrees downwards and tell the projector to calibrate its horizontal orientation, then turn it back to flat. Project against a white wall in a dark room. You’ll see (1 - sqrt(2)/2) = 30% of pixels (for a 45 degree tilt) will be wasted in total between top and bottom, and set to black, plus another 10% or so will be wasted at the sides for the black triangular cutouts beside the trapezoid. The original horizontal calibration was so bad that I would say more like 55% of pixels in the DLP were wasted, meaning that not only was the image warped (the aspect ratio was wrong and the keystone correction was wrong), but the total image brightness was 45% of the possible output for the lamp and DLP chip, and the image was smaller than it could be too.

That’s odd. There’s no way the accelerometer can have such a massive hardware offset, so my guess is that it wasn’t placed on a level surface during QA testing and that the projector retained the setting. I still haven’t received my PPM so I haven’t been able to verify the applied corrections myself yet. My goal is to have 1:1 pixel mapping avoiding any form of geometry correction (also maximizing the number of active pixels/mirrors in the process).