never thought this can ever happen!
As I am paranoid about precise calibration (to get the best results afterwards), I’d never expect this to be the case with me. By a pure luck – using PixInsight’s STF and STF max function – to check that calibration of single frames (that I perform always in MaxIm DL Version 5.10) is correct I have (with help of a friend of mine who actually discovered the issue) found out, that at least for last couple of years I had an issue in the master darks and therefore in the calibration process.
What was done a wrong way? Simply see following two images.
One would not believe that it makes a difference!
Checked with PixInsight’s PixelMath, when a single dark frame (instead of master dark frame) was used for both light frame calibration and flat field calibration, the issue observed was fixed. Strange! That meant I have constructed the master dark in a wrong way and this was not observable in MaxIm DL even after a huge (but linear) screen stretch. Thank’s to PixInsight’s STF I could have detected this minor (it’s very few ADUs) imperfection. From now on, I know that I must never save the individual master dark frame in MaxIm DL as IEEE float, but always construct it as median from Integer values. Positive outcome is that I can calibrate my last-year-taken data with a new set of master dark frames (hoping in better images then).