DSLR Calibration Frames. How many?? Best Practices??

[CRWPhoto]

I've had my share of cloudy rainy nights recently and have been thinking about calibration frames.  I can't seem to find a thread on this topic, although I am sure it has been discussed before.   I primarily use BatchPreProcessing however it's no problem for me to create master frames independently, therefore processing time and power usage is not a primary consideration.  These rainy nights give me plenty of time to waste on making the best calibration frames possible. Let's imagine for this thread that we live in a perfect world.  What's best for PI processing?

1) Dark frames.  I've been reading about dark frames and the concept of dark scaling and have come to the conclusion that rather than shoot a quantity of dark frames at the same exposure length as light frames that it is better and more efficient to create a master library of dark frames at various ambient temperatures with LONG exposures (longer than the longest light frames captured)... Then through the implementation of dark scaling, IP will scale the dark frames using noise evaluation to best calibrate with the lights.  For example:  If I plan on only shooting 2 minute subs, create a master dark library of 4 minute dark frames and simply dark scale. Regardless of whether I shoot 2 minute subs or 30 second subs, I can use dark scaling with the 4 minute darks for calibration.  (Any opinion AGAINST using this practice?  Also, how many should be captured?  I understand the math on diminishing returns, but as we will be making a master dark library, and only making each master once, what is the best qty to shoot for?  30 darks?  60 darks? 100 darks?

2) Flat Frames.  This one is pretty straight forward.  A master flat can be re-used over and over as long as the orientation of the optical train and the sensor are kept constant.  Of course, this will not help with dust bunnies or other variable anomalies, AND flats are so easy to take that it is not really a big deal to take them every time.  Either method is likely acceptable.  The big question:  How many? What is the perfect number for PI?  30? 100?

3) Bias frames.  I have read that for an excellent quality master bias frame to have the greatest impact when calibrating darks, flats and lights, that having MANY bias frames is ideal.  With 180 bias frames an excellent master bias can be created and that with 360, an almost perfect master bias can be created.  Is it worth taking 360 bias frames for PI to play with?

Thanks and fingers crossed for some clear skies!

-Chris

[NGC7789]

I would say the optimum duration of your darks is the longest subexposure you are likely to take. If you know you are not going to take subs longer than 2 minutes (due to light pollution for example) then there is no reason to take longer darks. You are right that you can then scale your master to any shorter subs. I would shoot darks on cloudy or moon filled nights and then have a script sort by temperature. Once I had accumulated enough for each temperature (I did 5 degree increments) I'd build my masters. I can't imagine there's much benefit to using more that 100 (which is what I use  ).

I shoot flats for every imaging session since I can't guarantee the whole system has the same orientation (I use a Borg refractor so I thread together tubes each time). Since there's really no cost I shoot 100 of these too.

Again for bias I shoot 100 because why not. I haven't tried superbias but as far as I understand that's the way to go if you want to get the most out of your bias frames.

[Geoff]

See Mike Schuster's posts in this thread http://pixinsight.com/forum/index.php?topic=7759.msg51474#msg51474 on how to calculate the number of flats needed.
Geoff

[astropixel]

Interested to see your results, particularly in relation to dark scaling. Matching darks to lights, exposure time and temperature, each session is laborious, but usually predictable.

[CRWPhoto]

Interested to see your results, particularly in relation to dark scaling. Matching darks to lights, exposure time and temperature, each session is laborious, but usually predictable.

Right now it's all just theory to me, and I hope to implement this type of thing over the next few weeks (clear nights *fingers crossed).  Jon Rista mentioned that dark frames shot with longer exposure than light frames will always be scaled down, resulting in a reduction in random noise, which may be more ideal than the "normal" method of matching dark frame exposure length to lights...  Not sure I fully understand the "why" with this, but it is appealing to me to be able to generate master darks at various temps with a specific long exposure rather than a whole mess of master darks at various temps AND exposure lengths.  makes for a MUCH more manageable dark frame library.   8)

Plus the concept of dark scaling makes sense to me from the viewpoint that within a specific imaging session the temp can drop 10 to 15 degrees (or more).  Plus, even if ambient temp remains stable, the sensor temp varies across the session.   This is really cool for those of us who are using uncooled DSLR's.