Originally posted by photoptimist That's a 4 pixel diameter circle for cameras with 5-micron pixels which is probably unnoticeable to all but persnickety pixel peepers. But a 4 pixel donut (probably with a 1-pixel ring around a 2 pixel hole) is much more noticeable.
Not sure that I am buying that when you take into account each pixel has an array around it as you describe and all are overlaying each other cancelling variation.
Thought I would have a go at demonstrating what I am thinking in the real world. The result is far too uncontrolled to make any serious conclusion but was interesting anyway.
The problem is there is no mirror lens without a central obstruction so I decided to use a refracting lens with an obstruction.
I took my A series 50mm f2.8 (chosen because it is sharp wide open) and put a 15mm black sticky dot on it's appr.20mm front element. That calculates out at anything with a dof greater than that of a f3.3 aperture is obscured. Obviously this is a much greater obstruction than your average Mirror lens.
Remember that the outer edge of a lens has a lower optical performance so you can expect the image to be not quite as sharp as the clear lens with its sharper internal sections enhancing the image. (That comment in itself reinforces my original argument too doesn't it?)
The top half of the crop is totally about bokeh and left in there to show how well this dot has replicated a mirror lens.
And a bokeh donut is visible in that background bokeh near the bottom.
I focused on the vertical frondlet and you can see some degradation in the dot view. (I didn't change focus after applying the dot.)
The frondlets bottom left and the stem interest me as they are that "just out of focus" area that I was hoping to see a difference.
And I believe they have softened just a little more in the dot image.
But a far too inaccurate a setup to be a definitive result.
Of course this is not an exposure comparison and I have brightened to dot image to match the other.