I just want to verify that I am not missing some setting somewhere. My original lights are .CR2 and about 8M each. My lights after conversion are about 20M each. But the pre-processed lights (w/ master-dark and master-flat) is about 120M each.
Is this expected? (trying to process about a 1000 lights and didn’t expect this scale of increase in pre-processing size.)
Yes, with the recent versions of siril, the default is to use 32 bits per channel per pixel instead of 16, to get better precision and handle negative values of preprocessing in a better way too.
You can go back to using 16 bits either by enabling it in the preferences in the FITS options section or by using SER sequences which do not support 32 bits per channel.
Also, the preprocessed files are probably demosaiced in the same operation, which makes them 3 times bigger.