See, for example, Very Long Baseline Telescopes.
sorry but aren“t that radio telescopes??
Yes, but there are now optical telescopes using the same principle.
The fact is that if you take a large lens that gives you high resolution because of its large size,
you can paint most of it black and leave odd spots of transparent glass around the edges and, I guess, some dotted around inside, and still get that high resolution.
That's why VLBT telescopes exist; there'd be no point in them otherwise.
It's just that your light grasp becomes rubbish when you paint most of your lens black!
So you have a high resolution, but a RUBBISH T-stop because most of your lens/mirror/desert isn't contributing
signal.
Having said that, I am sure there are Fourier "consequences" of only having odd dots on your lens/mirror/desert contributing. I expect the COC "shape" depends on the FT of the pattern of the dots you have chosen. Or something. I'm straying into hand-waving semi-guesswork here; it's been over 20 years since I did Fourier signal theory at uni.
Having said that, I do have a book entitled "Atlas of Optical Transforms" which is a whole book full of 2-D patterns in real space on one page, and the optical Fourier Transform 2-D on the page opposite, so you can compare them and see how one is related to the other.
Then it goes on to show effects of masking out part of a Fourier transform pattern and how the masking affects the reconstituted-from-Fourier pattern back in real space. So you get low- or high-pass spatial filtering.
Fascinating reading, it is.