I took a closer look at SGP’s FITS files from my Canon EOS 600D (14-bit) and found that SGP scales the original values in the following manner:
I = I0 * 4 - 1
(The subtraction will not be executed if I0 is zero.)
This means that the scaling does not destroy any data, and we can withdraw the scaling in PixInsight by the following PixelMath calculation:
($T*2^16 + 1)/2^18
(‘Rescale result’ must be unchecked.)
I did the PixelMath on one of SGP’s FITS files (that was saved as CR2 file as well) and compared the result with the corresponding CR2 file that was raw developed with dcraw v9.27, parameters: -4 -o 0 -r 1 1 1 1 -D -t 0 -k 0 -H 1 -T
The results are identical. In this way you can convert light frames from a camera with 14-bit ADC that are saved as FITS files by SGP so that they are compatible with calibration frames that are saved as CR2 files. Up to now I saved FITS and CR2 files. In future I will save my light frames from SGP as FITS files only.
I guess that files from a camera with 12-bit ADC can be converted analogously.