With programs like Starry Landscape Stacker or Sequator image stacking for noise reduction has become very easy to do.
The one question I get is how many images should i stack?
This question can actually be answered analytically with math. If you recall your stats class, the variance of the sum of n variables is the sum of their covariances. Therefore…
Ok just kidding. I won’t go into the math. But if you want to dig further look at the Wikipedia article on variance and assume that the photons have a Poisson distribution.
The way the math works out is that if you have a stack of N photos and average them, the resulting noise is 1/sqrt(N) of the original noise.
In other words, if you stack 4 photos, the noise will be reduced by 50%. Stacking 16 photos reduces noise by 75%. Here’s a table summarizing the noise reduction for a few values:
Looking at the results, there’s clearly diminishing returns. In practice, I usually take anywhere from 10-20 images when stacking. I may not use them all, but I have them available as an option.