[frank]

I also made a test with the camera rotated 90 degree to the other side (equivalently, the computer display is rotated by 180 degree relative to the above test). （But afterward I found that the right boundary of the template image was a little bit farther from the camera than the left boundary.） The results are shown below. If these tests with camera rotated makes sense, then perhaps it can be concluded that the brightness/contrast problem of the computer display at least partly contributes to the nonasymetric pattern of the blur.