Hey Andrew,



This can be confusing, because the size of the file is defined differently in Partially Coherent Image Analysis than it is for the other bitmap based features. That's because we expect you to want to look at image quality at the scale of the PSF or maybe 10xPSF, so that you can see the effects of partial coherence as the image enters diffraction-limited performance. So, we wanted to make it easy to work in image-space rather than object space as per the other bitmap features.



In the double Gauss 28 degree sample file, I set up the PCIA like so:







In this case we use the letterF.IMA file which is a 7x7 bitmap. With no oversampling or zero padding (as done here) you can see that you get exactly what was stated: (1) a file size of 1 gives (2) an image of the letter F which is exactly 1 mm across, or +/-0.5 mm. This is 'The full width in lens units of the region defined by the file measured in image space', as stated.



Now note setting (3), the Field setting. This is where the IMA file is centered, in this case it is centered on field point 1 which is the center of the field of view in this case. If you select field point 2, you will be an image of size 1mm centered on that field.



So to calculate the real size in object space, you'll need to account for both the magnification and distortion of the lens by tracing real rays that land at the top and bottom of the image and getting their coordinates on the object surface. However, the image is likely very small so that you can see the effects of the spatial coherence function, and so distortion is unlikely to be an issue. It's best to remember that we expect the image size to be small, or the order of 10*PSF or so when using this feature.