Yes, thanks! I think I understand all of those concepts but it’ll probably take
me awhile to figure out how to actually do all of those steps. Maybe I’ll have
it figured out by the Messier Marathon in the spring. :-) In the meantime, I’ll
experiment a bit this weekend at the AASP and hope to be lucky.
On Oct 3, 2018, at 3:29 PM, Greg Schwimer <schwim@xxxxxxxxxxx> wrote:
This may be helpful. I can't avow to its accuracy.
A rule of thumb that's worked for me is to take a set of darks at each
exposure I might want to use. Stack them without bias calibration. Measure
the median signal of each. This is the "floor" of noise at that exposure
level and you need to get the background of your image above it. You want to
"swamp" this noise floor with signal, so the median of your subs should be
higher than that of the darks.
An approach for determining how high above that dark I've seen used
frequently is to subtract a master bias from the dark and take the median of
the result. Triple this value and add it to the bias median value. Some say
this is the minimum you should go for. I admit this can be tough on some
equipment (e.g. slow optics and/or narrowband). Once you have that value as a
starting point, take a few sample images and measure their median.
All cameras are different, and narrowband vs. RGB vs. bayer filters all have
an impact due to the amount of signal that the filter allows to the sensor.
Dark skies vs. light polluted skies will play a significant role in the
results as does the brightness and size of the objects you're imaging as
As for number of exposures, I believe the general rule of thumb is that SNR
doubles with the square of the exposure count. That is, 4 frames has double
the SNR of 2, 8 has double the SNR of 4, and 16 the double of 8. I generally
go for 16 and call it done unless I see something very faint I'm after, but
again there is a point of diminishing returns. You can see this play out
quite easily by comparing stacks of multiple sizes side by side.
Hope this helps!