Quote:
Originally Posted by Paul Haese
The general rule of thumb that I work to is that back ground ADU has to be around 1000 to overcome noise in a certain amount of subs.
|
I agree, that's the basic principle behind the Starizona calculations. It then also follows that the optimal sub exposure length is different for each filter.
I calculated the optimal for my L and then experimented with the length of R,G and B until they displayed the same background ADU count as the optimal L. The result is four different sub exposure times and that's what I use now.
Quote:
Originally Posted by gregbradley
Hi Rolf,
Isn't the deep mag 24 a result of the long total hours rather than the number of subs?
|
Yes the total time was indeed longer. 14 hours versus the 8-9 hours I had been doing on other images. But there is not an order of magnitude difference in those times.
Interestingly, I could clearly see from the raw data that stacking say 20x5mins gave superior S/N compared to 10x10mins so that's why I was convinced shorter subs was the way to go. It may even be better to stack 40x2.5mins, I just didn't try
Quote:
Originally Posted by gregbradley
Each time you do a sub the signal has to rise above the noise floor to be added to the signal. So faint details may in fact be lost as they are never recorded in the first place in short exposures. Adding exposures doesn't work unless the signal rises above the noise floor otherwise the faint signals are lost in the noise.
|
I'll say the signal is always recorded, unless the exposure is so short that no photons actually reached the chip (When we are talking exposures of a minute or more then that's probably unlikely).
The value of each pixel is the signal
plus the noise. In the final stacked image the faintest details are typically only a few ADU counts higher than the background. It's only when the signal itself approaches the square of the averaged noise background that it gets lost. But the signal is always there in every sub frame, it's just a matter of having enough subs to smooth the noise enough to reveal the signal.