I was wondering if anybody has got a good strategy to deal with entirely masked rows or columns when using profoundPixelCorrelation to do correlation plots? This happens quite a lot for my cutouts if they go over edge and just running them through profoundPixelCorrelaiton like this throws an error ("no complete element pairs"). Cutting off the masked rows/columns obviously is an easy solution (at least if they're at the edge of the image and don't go right through it), but then the image isn't square anymore and I'm not sure what profoundPixelCorrelation does in that case (the FFT seems to be working fine; while the correlation plot and cortab look to be cut off somehow at the shorter dimension - and it gives a warning as well).
Ah, I see. The image needs to heavily masked (but not necessarily entirely) for this to be an issue. I've fixed by changing the default NA handling behaviour inside the function. Try the newest version on GitHub.