yes, there is a thread on this that i probably started. juan explained how to set up the fits reader to properly read in maxim DL f32 fits files. bottom line is that maxim is using the f32 format to store i16 data, so the range of values in the file are [0.0 - 65535.0] rather that [0.0 - 1.0] which is what pixinsight uses.
that's problem #1. fixing that makes your bias and darks work okay.
problem #2 is that the "dynamic range" (for lack of a better word) of the flats seems to be way too high. where the lights seem to want to be divided by 1.0 at the center to maybe 0.98 at the edges, the flat's illumination at the edges is much, much less and therefore when you divide the light by the flat the vignetting/dust is WAY over compensated.
i asked steve to send me a DVD with the flat subexposures under the theory that maxim DL must have done something very funny to the flats when integrating them... but he ignored me
