Introduction and calibration question

[Duncan]

First, since this is my first post here (hi!) I thought I should include a brief introduction:

I have been imaging for about 12 years, starting with a DSLR attached to a 6" Newtonian, and iterating my setup many times. Currently I have two main scopes, a Tak FSQ-106 and a Celestron Edge-HD 11, plus a QSI583wsg, mounted on an Astro-Physics Mach1. In the past, I've used numerous image processing and capture software systems, previously ImagesPlus (now a long time ago with the DSLR) and for the past several years Maxim DL. I'm still intending to use Maxim for capture and control, but I've recently started using PixInsight with the hopes of getting finer control over the process and some extra processing features.

I'm just getting started reprocessing some data from last summer, running the latest rev of PixInsight on a desktop machine I built for the purpose (with a Core i7-4790 & 8GB of RAM).

Anyway - first of many questions:

I've read (and watched) many tutorials, but wanted to check that what I'm doing with calibration is sensible; I'm doing something a little different to any of the tutorials that I've seen.

What I typically do is take flat-darks, ie a set of darks with the same exposure time as my flats (2 seconds in this case).

I start by building master bias frames by just integrating a set of 50 bias frames, and then a master flat-dark and a master dark the same way, and integrating each.

Then for each filter:

I then calibrate the flats using the bias and flat-dark. I have the "calibrate" checkboxes for the master bias unchecked, but for master dark (which is the flat-dark) checked, because I didn't previously calibrate the flat-dark frames individually. From that, I build a master flat with integration.

Finally for light calibration I use the bias, master flat and master dark frames, with:
calibration unchecked for master bias
unchecked for master flat
checked for master dark, because no calibration was applied to individual dark frames.

Is this correct? Any comments or suggestions much appreciated... 

[cwiede]

Hi Duncan,
I'm a newbie myself, but I also use flatdarks. I build masterflatdarks and masterdarks just by averaging (or by making the median) - no bias correction! Also when I calibrate the flats I don't use a masterbias. Each flat contains the flat signal, the dark current and the bias. The masterflatdark contains the dark current and the bias. So if I subtract the masterflatdark from each flat, the flat signal is the only thing that remains. Same thing with the calibration of the lights. If I subtract the masterdark I get rid of the dark current and the bias. The light signal is the only thing that remains.
I never take bias frames. I think they are only needed if you want to create a new masterdark using another masterdark aquired with different exposure time. In this case you would subtract the masterbias, rescale the dark current and add the masterbias again.
Hope this is correct...
Best regards
Christof

[pfile]

with respect to dark scaling, you don't have to make new dark masters. in ImageCalibration, just turn on calibration and optimization of your master dark, assuming it was composed of non-calibrated dark subs. if you have a calibrated master dark, then just turn on optimization only.

with my setup i have found that subtracting a master bias from the flats results in good flat calibration. i suppose if you are working with a camera with a lot of dark current, or you can not get the CCD temperature way down, then you may need flat-darks. maybe narrowband lightbox flats could be so long that a master flat dark is warranted in my system, but i usually do sky flats and they are not overly long.

rob

[Duncan]

thanks for the replies - much appreciated.

Christof,
thanks - that helps my understanding. I'm still a little bit in the mode of trying to work out what everything does, and so it's very helpful just to check how everybody does things.

Rob,
thanks also - that makes sense.

I tried an experiment to see what my flat-darks were actually doing by running a calibration on my master flat-dark with my master bias. What is left is almost zero, so it looks like there is almost no dark current signal in the flat-darks. They're only 2 second exposures; I'm using some data from a dark site with LRGB filters. I do also have a narrowband set that I use at home (in my Bortle 10 skies  ) but my flat box can be turned up quite bright for that case.

I also looked at the noise in the "calibrated" flat-dark and it looks like exactly what you would expect from adding two independent noise sources (it's quite a tiny amount of noise).

I guess the conclusion is that with my setup, there's really no point in using flat-darks as well as bias; doing so isn't fixing anything and is just introducing a small amount of additional noise.