Hi,

I am wondering about the inner workings of the Image Simulation feature of OpticStudio. Having read this discussion and the built-in help accessed from the info button in the Image Simulation settings, it seems that this function computes the PSFs at several points as specified by the user, and then interpolates between these computed PSFs to approximate a PSF for every point on the object. Then, each point on the object is convolved with the appropriate PSF to produce the simulated image (I skipped over additional processing like scaling/rotation that is mentioned, that’s not what I’m interested in here). My question is: how exactly is the interpolation between these PSFs performed? Does Zemax do principal component analysis of some sort, and interpolate those components? This does not seem to be described anywhere.

Particularly in my optical system, the PSF at the edge of the field of view may be crescent-shaped, and the system is rotationally symmetrical so at any constant radius from the optical axis the crescent would vary with angle only by rotation. Directly interpolating between two sampled PSFs at the same radius from the center but at a different angle would logically seem like it would yield (incorrectly) a cross shape of some sort; but OpticStudio manages to avoid this mistake somehow when I try it! The papers that I’ve found on the topic of interpolating between PSFs have all been relatively recent, and mostly would not be able to avoid this mistake the way that OpticStudio does (the fact that OpticStudio does not make this mistake leads me to believe that principal component analysis is not used).

Is the algorithm itself documented somewhere? I am interested in using this as a justification for my choice of PSF sampling based on my optical system properties (for instance, it may be advantageous to set PSF-X Points to something high but keep PSF-Y Points = 1 if the interpolation method handles rotational symmetry well).

Thanks for the clarification,

Daniel