DSO Astrophotography Stack Size

In my DSO imaging I usually gather around 100 frames of each target and stack them. I don't have any good reason for using this number - it is based of empirical evidence and the experience of others.

I decided to experment with some of my AVIs to see how increasing the stack size effects the final image. Performing such an experiment is difficult. When we post process we spend a lot of time working on the final image - but to show the differences in stack size the processing must be kept similiar to be fair. A long post processing will be hard to replicate.

I have taken AVIs of NGC6503 and M63 and used registax to make tiff files showing stacks of 5, 10, 20, 40... frames. Each was cropped identically in photoshop. The only processing was done using the levels control instead of the usual curves control. I just moved the histogram sliders to stretch each channel across the full dynamic range. This does not make for a pretty picture, but illustrates the point. No dark frame was used.

NGC 6503 M63
5 frames
10 frames
20 frames
40 frames
60 frames
80 frames
100 frames
116 frames

NGC 6503

As you may know, them more frames you stack the less noise in the finished image. However, it is a square rule - one of dimishing returns. Looking below at the raw frame, one can hardly believe the finshed image came from it.

1 frame - just a noisey smudge.
5 frames - still a noisey smudge, but some of the struture in the final image is just visible
10 frames - structure visible, background very noisey.
20-60 frames - detail increases slowly, but background noise drops
80-116 frames - hard to tell the difference with this processing.

In conclusion, I would say that the minimum number of frames of a bright galaxy such as this is about 40. More than 80 frames does not give a large increase in quality.


The coarse levels method used to ensure similiar processing has burnt out the core of the galaxy. Again, the finished image does not seem to belong to the raw frame.

40 frames once more seems the magic cutoff where the gain in image quality starts to become small.

It should be noted that these results are the possibly biased by the crude stetching done to process each frame. The next step would be to process the 60-80 frame images to the best of my ability, and see what differences are visible.

Page last updated 2005-09-14