I've been using PixInsight for about a year off and on, as the skies allow me to get data, and my primary method of acquisition has been a SX H694C. Since it has very low noise, I would generally just take flats, flat subtract each frame, then debayer, align and integrate. Sometimes I would debayer first, then flat subtract, but occasionally I would get a problem where flats were just over-corrected to an extreme case. Recently, I bought a Canon 6D for imaging as well, and with the much larger sensor, flats are even more important. Now the problem crops up all the time (maybe the SX was small enough that the magnitude of overcorrection was small enough to not bother me) and is very severe. It seems to be linked to the magnitude of the vignetting. I've tried making the master flat with both additive and multiplicative integration and I get the same results. I use the ImageCalibration tool for the flat subtraction, and I've tried both with and without checking 'Calibrate' (though, I use no other calibration steps listed).
Acquisition is either done with Nebulosity or BackyardEOS, and all raw images are then dumped into PixInsight for processing. I've not yet tried to do flat subtraction with another tool.
I'm sure people will state that I should be subtracting bias and darks, but honestly, it only seems to make the noise worse. I get much better results without it. I've tried doing it all manually as well as the BatchProcessing tool. That's a whole 'nother topic. So my question is, if my calibration consists only of taking flats, how do I do this properly so I don't get over-corrected images with flat subtraction?
