›

Quantities of calibration frames?

Darren Jehan
in PixInsight
So, putting together some basic tutorials for beginners in our astro-society and we get to the topics of calibration frames.
It occurred to me that I take an arbitrary number of frames for the different types, but I have not actually applied any logic to those numbers. Currently I use 101 bias, 17 darks, 21 flats (plus flat darks when using my CMOS camera).

I would like to - if nothing than for my own understanding - work out if these are actually adequate or not!

The question is, how do I go about determining what those quantities might be?
What am I measuring, and with what?

Any insights / guidance appreciated!
(I haven't seen a specific video on this topic, so apologies if there is actually one!)


Thanks all!

Comments

  • BlockHeadBlockHead
    The answer to your question is technical. It depends on the camera characteristics (such as readnoise) and statistical noise reduction of all noise terms. Basically you do not want your calibration frames to *add* more noise then the smallest uncertainty (which affects your faint limit in data) you are willing to accept.

    Berry and Burnell explain this well for amateurs in their Book "The Astronomical Handbook of Image Processing." You will find that roughly.... 50 biases is probably good enough...and in essence...you can never take too many dark frames. lol

    Here is my non-technical take on things.

    1. 50 Biases is probably good enough.
    2. You need probably *more* darks... but it depends on your total integration time, sensor characteristics...etc...read the book. I would be happy with 30-40 darks... and keep in mind my darks are 20-30 minutes each!
    3. Less flats are probably fine. Assuming you do not have stars in your flats (no rejection issues)- then flats have very low uncertainty in the sense you are just minimizing the poisson noise and not readnoise... flats are BRIGHT. Thus, to achieve a certain noise calibration level, you just need to take the sqrt(N) flats to achieve it. This is what astronomers do when doing photometry to measure small variations in brightness...they need to calibrate the pixel response to this or better level. BUT...if you are not doing photometry and just pretty pictures- you cannot see this level of precision beyond a certain point. 11 frames per filter is likely quite good. More frames certainly will not hurt. Why 11? There is a minimum number just for plain old statistical rejection. (Theory says it is 7 measurements... but in practice I use 15... and 11 is just fine).

    -the Blockhead
  • Darren Jehan
    Thanks Adam, I'll go hunt out that book!

    Daylight and cloudy nights exist to obviously take more darks!


  • BlockHeadBlockHead
    I only agree with the second part of the sentence....
    Too many people have light leaks. 
    -the Blockhead
Sign In or Register to comment.