Hi Jose,
the normalization used for flats, bias and darks before combining them.
Light frames:
output normalization = additive + scaling
rejection normalization = scale + zero offset
scale estimator = iterative k-sigma / biweight midvariance (IKSS)
Flat frames:
output normalization = multiplicative
rejection normalization = equalize fluxes
scale estimator = iterative k-sigma / biweight midvariance (IKSS)
Bias and dark frames:
output normalization = none
rejection normalization = none
scale estimator = median absolute deviation from the median (MAD)
if the darks and flats are bias calibrated before combining them
The generated master dark frames include the bias signal. If you use externally generated master dark frames, they must include the bias signal.
The generated master flat frames do not include the bias signal, since each individual flat frame is calibrated independently, and hence bias-subtracted. If you use externally generated master flat frames, they must not include the bias signal.
Let me know if you need more information.