Requesting comments or suggestions on the calibration workflow that I’m using right now. I’ve tried a few different process variants, based on reading a number of different threads here and some tutorials in various places. I didn’t keep notes on where I got everything from, unfortunately. What I have right now seems to work very well; I get some very clean frames with no apparent residual vignetting. Nonetheless, I’d really appreciate any commentary on my current methodology and whether there are any improvements I should consider.
The basic process is this:
Integrate bias sets
Integrate dark sets
Integrate flat sets
(I’m applying no calibration to darks or flats prior to integration)
Calibrate lights using the masters created above with (mostly) all calibration/optimization checkboxes checked.
My light frames use either 1x1 binning (for L) or 2x2 binning (for RGB) so I have 2 sets of each of bias and darks, and the flats for each filter have the corresponding binning. My primary camera is a QSI583 which has a very stable regulated TEC, and I standardize the temperature set point at -10C which I can reliably hit under all conditions.
For integration I use Average combining with no weighting and no normalization. For each set, I do a first run with no pixel exclusion and save it. I then do a second run with an exclusion algorithm. Winsorized clipping seems to work well (I have usually 50 frames in each set). The dark frames are 10 minutes each, so I can get a pretty large number of cosmic ray artifacts. I take a pretty good look at the high exclusion image after integration to see that I’m getting rid of the cosmic ray artifacts and as little as possible else. I’ve also been using the blink tool to compare the excluded version with the saved non-excluded first run, which seems to give a nice indication of whether I have the exclusion set up right. A sigma of 4.0 seems to work most of the time, but I’ve used 3.5 on some sets.
For the calibration of lights, I set calibration off for the bias frames (there’s no overscan region set). I set calibration and optimization on for darks, and calibration on for flats.
The result of all this looks really good (I do subsequently run cosmetic correction on lights also).
I do usually see a warning that no dark subtraction was performed on the flats due to no correlation of noise evaluation. I’m assuming this is ok; the flats are taken using a flat box as a source which is quite bright and the exposures therefore quite short, so I’m assuming that any dark noise type artifacts are just being overwhelmed by shot noise so there’s nothing useful to correct for with dark subtraction.
Long post, I know, but does this seem sensible? Anything I should consider changing? Any feedback/suggestions/constructive criticism very much appreciated…
Duncan
