In the beginning of my PI experience, I used the BPP script for calibration (and sometimes stacking). Recently, I've been trying to improve my calibration, and have done things manually. I've noticed an error I made, that might come in handy for other users:
I currently use an Orion Starshoot Pro v1 OSC camera. When i measure Median (using Statistics) I measure pretty odd values on my Bias/Dark stacks. A recent bias stack measures K=0.015 and the dark stack measures K=0.014. With an automatic STF, they look rather different though. The bias stack is way smother than the darks (100 bias vs 60 darks@300s). The frames were shot outside, but since the camera has unregulated cooling, the ambient temperature might have varied up to 5 deg Celsius.
The normal procedure for calibrating the dark stack is to subtract the bias stack:
Calibrated Darks = Darks - Bias
However the subtraction will give negative results for the calibrated darks, and for some bizarre reason PI truncates those values to zero. That means, that you end up with a calibrated dark stack consisting of hot pixels, and all dark noise is set to zero! Using such a calibrated dark stack, to calibrate your lights, won't remove any dark noise at all...
Sometimes the measured K values of the stacks are very close, meaning that some dark noise will be positive values after calibration, but all the negative values are still truncated, rendering the calibrated dark stack pretty useless.
The solution is to enable pedistal while using ImageCalibration in PI. This adds a small value to the calibration, so you get rid of the negative values, and the truncation. At least on my images, the difference was amazing. I've previously seen K values that were very close on DSLRs as well, so be aware if you calibrate your darks (good for scaling them).
Morten
