Kevin_A do you feel that there is much advantage taking darks with the 2600 series along with flats and flat darks?
I do it because that is what the math says about sensor errors (albeit small for something like the 2600).
Now, if you are concered about the dark frames adding noise to the result, I don't think you need to be. If you make up the dark frame (basically, same exposure parameters as the light frame) with 1 single dark frame, you will be decreasing the SNR by 3 dB (i.e., you are adding the same amount of noise but no data -- the lens is capped). I.e., the total noise is sqrt( L2 + D2 ) where L is RMS of the Light frame and D is the RMS of the Master Dark frame. Since L=D (D is noise from a single Dark frame, remember?), we have total noise = sqrt(L2 + L2) -- thus we are doubling the total noise after adding the Master Dark to the LIght, and thus a 3 dB increase in noise. This then translates to 3 dB worse signal to Noise ratio. Not so good -- but remember, the Master Dark is made of only 1 single Dark frame.
OK, once you grok the above, now consider making the Master Dark from 2 independent dark frame (again, just two frames), The Master Dark now has a noise that is 3 dB less than the noise from the 1-frame Master Dark. So, now you are adding 3 dB less noise to your LIght frame. The total noise from the 2-frame Master Dark and the Light Frame is now sqrt( L2 + (L2)/2 ) = L*sqrt(1.5). Ha ha, noise power (variance), the inside of the sqrt, has gone up by just 1.5 now (not 2).
log10( 1.5 ) = 0.176, and thus 10.log10( 1.5 ) = 1.76 dB. Yippee, the noise has ony increased by 1.76 dB now, by using 2 Dark Frames instead of 1 Dark Frame.
Now, if you double the number of Dark frames again, to 4, the second argument in the sqrt halves again. So, a Master Dark with 4 Dark Frames will create a total noise of L*sqrt( 1.25 ) or 0.96 dB. Your final SNR is now worse only by less than 1 dB now.
OK, lets take 16 Dark frames. The sqrt is now sqrt( 1.0625 ) or 0.263 dB worse.
With 32 dark frames, the SNR worsens by only 0.133 dB. With 64 dark frames, the SNR worsens by 0.0673 dB.
I doubt very much that the human eye could tell the difference of 0.1 dB in SNR. So with just 50 dark frames, you can be pretty sure that applying the Master Dark would not ruin your final image.
I usually do between 20 and 30 cal frames of darks, flats and flat darks to create masters.
I still would recommend more than 20 (0.21 dB SNR degradation), though, since how often do you take Dark frames for the library anyway. So you can knock yourself out, and take 100 dark frames (I use between 100 and 200, but I am an OCD engineer :-) and there is no way Superman can telll the difference. At the same time, the ASI2600 dark current noise is likely to be more than +0.1 dB, in which case applying a Master Dark will improve the final SNR, not worsen it.
Once you create a Master Dark, the actualy processing time in the future is a constant, whether the Master darkk started with 10 Dark frames or 1000 Dark frames.
Whether it is worth it is up to you, but I typically start up AstroPixelProcessor and walk away. Plus, my Mac Studio Ultra has 128 GB of RAM and I give 96 GB to AstroPixelProcessor -- not typo - 128 Gigabytes of RAM; I maxed out the memory that the Studio can support when I bought it, since the plan is to use the computer for 5 years -- it has been 2 years already. I give AstroPixelProcessor all 20 cores of the CPU. So, I have no idea how much extra time it takes with or without applying Dark frames, but we are probably talking seconds extra. (The extra money spent on maxing out the Mac Studio has probably paid for itself in time saved, already, without having to upgrade the computer every year -- just like extra money spent on Takahashis :-).
Chen
