Rayleigh's criterion doesn't hard limit your ability to aim-- so long as you have a good prior what the image plane projection of the target is. To give a cont

Rayleigh's criterion doesn't hard limit your ability to aim-- so long as you have a good prior what the image plane projection of the target is. To give a contrived example. Say we have a function f(x) == a*exp(-(x-m)^2). I evaluate f(x) at 1 and 2 and tell you that the evaluations are f(1)=0.27093242847450331643480903517 and f(2)=1.1666240719274935885932696286. From that, you can tell me the amplitude (a) and peak position (m) with on the order of 100 bits of precision (in this example). This is contrived because its noiseless and you know the system response perfectly, but I think it demonstrates that an alignment under assumptions need not be limited by rayleigh's criterion.