Shack Hartmann lenslet array design

I'm going to guess you're not talking about a plenoptic camera but rather a Wavefront Sensor.

Given that I worked for a company called Adaptive Optics Associates for many years, I'm painfully familiar with Hartmann sensors. In brief (and you should read the references), your tradeoff is between maximum tilt dynamic range, or wavefront tilt, per subaperture, and response linearity & precision.

You get to choose the focal length of your lenslet, which defines the Airy disc diameter. The Airy radius corresponds to one wave of tilt.

Next, compare that with the size of each pixel. In the simplest case of a 2x2 "quad cell", your spot will be completely on one side of the null point at 1 wave of tilt, so beyond that you know only the direction, not magnitude of tilt.

When you "assign" a larger number of pixels per lenslet, say 5x5 pixels, then you can observe the focussed spot's motion across all the pixel boundaries. The raw output will have "wiggles" but these can be calibrated out of the system.

I would like to post a link to the famous MIT-LL "Adaptive Optics Coloring Book" but it seems to have disappeared from the 'net.