During one of my recent projects, the Heart Nebula, I decided I would stack some images at different intervals and report here on my results. I used my SII filter for this test. SII usually will not have anywhere near the signal you'd get from the hydrogen alpha filter. As a matter of fact, I did not use any noise reduction at all on my hydrogen alpha data of this region. I thought SII would be a good filter to use for this test, because we can better see the awful noise and artifacts.
So first of all, why stack multiple images? Stacking images can actually be used in any setting that isnt changing to increase resolution and decrease noise. Or in other words, it will increase your signal-to-noise ratio (SNR) and increases the dynamic range. Astrophotography happens to be perfectly suited in this manner. Deep space objects (DSO) are virtually permanent and remain static for long periods of time.
You also want to stack multiple images because each exposure will inherently have cosmic rays, satellite trails, airplanes, and noise in both hot and cold pixels that stacking algorithms like sigma clipping will correct.
Lets see how stacking more and more images improves the final data we have to work with. Each of these are a stack that has had a Screen Transfer Function (STF) applied in PixInsight so that we can actually see what we have. Without the STF applied, the images are very dark and we would not be able to tell anything. You can see how the images gradually increase in SNR and ultimately become much easier to work with.
Each of these stacks are from images that have been binned 2X2 and 2X drizzle has been applied in the stacking to gain back most of the resolution. Please take note of the black specks (cold pixels) in each image and how they are gradually reduced and ultimately eliminated. These are center crops at 100% and each image stacked is a 900 second exposure.
 |
| 5 subs |
 |
| 9 subs |
 |
| 12 subs |
 |
| 15 subs |
 |
| 23 subs |
 |
| 28 subs |
 |
| 40 subs |
So, how many would be considered good enough to you? Some would say 15-20 subs would be enough and would not want to spend 10 hours on one filter. This is perfectly fine. As a matter of fact, I'm working on one right now that I will likely call good enough after 20 subs with the SII and OIII filters. But you can clearly see, if you look close enough, that adding more and more can only help the final image. There is, however, a point of diminished returns. I feel that 40 works well for me when binning and drizzling the SII and OIII data. I shoot my H-alpha at full resolution (1X1) using 20 minute subs and find 8-12 hrs works very well. When shooting this long I sometimes do not even need to use noise reduction.
The above center crops have had no processing at all. Here is a crop of the final SII after processing.
 |
Final crop of the SII
|
Here is a center crop of the final image using the hubble palette to assign the color channels. SII for red, Ha for green, and OIII for blue.
 |
| Final SHO, center crop |
And finally, here is the final full image. The full screen image can be seen
here, and by clicking on the icon at top right it can be expanded out to full resolution.
I hope this has shown some beginner and intermediate astrophotogaphers why it is important to get enough subs to make a nice image. It is ultimately left to the astro imager to decide for themselves how many is enough, and there is no right or wrong way to do this.
Happy imaging everyone, and happy Thanksgiving!
