But Juan ignored the report
I don't ignore bug reports---what's the point in doing that?
It's simply that I overlooked your report for some reason (I didn't answered you, which supports this).
I can confirm the bug. I can see where's the problem in the code. I'll fix it ASAP, although it actually has no practical relevance. In addition, I'll remove the 'same as target', '8-bit' and '16-bit' options because they make no sense.
So Sander, stop waiting for "there is no bug here"!
Also, I think ImageCalibration (I have code for Version 1.1.1.0077 only) is not compatible with 64bit images, because all MasterFiles limited to FImage(32float).
That's true. For example, in line 1690 of ImageCalibrationInstance.cpp you can see these declarations: Image* bias = 0;
Image* dark = 0, * optimizingDark = 0;
Image* flat = 0;
So you're right, all calibration frames are loaded as 32-bit floating point images irrespective of the original data type. The same happens for all target frames. This isn't a bug, but a simple design pattern: Actually, working with 64-bit images makes no sense at all for a calibration routine. Supporting them would complicate the code (and would duplicate its binary size) without any justification, from both theoretical and practical considerations. All intermediate calculations are always performed in double precision, but this is a different thing.