Monday, April 21, 2014

Autostakkert / Avistack - a casual comparison

In this post I will compare the output from two of the best programs for sorting, aligning and stacking numerous exposures: Autostakkert v2.3 and Avistack v2.0. The subject is solar H-alpha images. The latest Avistack version is from July 2010 so it is not exactly undergoing a hectic development! Autostakkert is newer and is in a more dynamic development phase with minor updates occurring many times pr. year.

I did not spend a lot of time tweaking the processing parameters with Autostakkert - instead I deliberately did a 'quick run' using most settings at their default values. Avistack I know quite well and have been using for solar image processing for the past two years. During this time I have tweaked the settings to best suit my data. Hence, in a way the comparison is not 'fair' - that's why I called this a 'casual comparison'! See the cropped images below for the results, figure 1 is a prominence and figure 2 is an active region around a sunspot.

Figure 1 was made using 1500 frames from a Coronado filter on a 70mm f/7 refractor. The best 60 were used for the final image which then had wavelets applied to sharpen it (same settings for both images of course).

Figure 2 was made from 700 frames taken with a Daystar filter on a 6-inch f/6 refractor. Again, the 60 best frames were used but this time no wavelet processing was applied.

Fig. 1: comparison on a prominence.

Fig. 2: comparison on a sunspot.
The results are not dramatically different; both Autostakkert and Avistack do a very good job. However, the Autostakkert result is generally slightly better with just a little more 'omph' to it. Details of the settings are given at the end of this post - notice that Avistack has many more parameters to set. I have tried to choose settings for the programs that were similar to each other, but of course I cannot rule out that some Avistack settings exist that would allow matching or even surpassing Autostakkert. However, I do know Avistack quite well and have tweaked its parameters extensively in the past - I find that the results do not depend super critically on the chosen parameter settings.

In conclusion: Autostakkert gives a slightly better result and has much fewer parameters that require setting - a clear advantage, in my opinion. Some will balk at this 'lack of control' but sometimes there are just too many things to adjust and you can never be sure that the correct settings have been found. I find that Autostakkert is simple to use and works very, very well 'right of out the box'.

Some of the parameters used for the comparisons:

Autostakkert: image stabilization: surface, noise robust: 3, AP size 50, HQ refine: on, Drizzle: off, number of frames to stack: 60.

Avistack: quality analysis: on, frame cutoff: 60, area radius: 35, search radius: 28, alignment type: surface, ref. point min. distance: 13, structure threshold: 44, lower cutoff: 0.23, noise reduction: 0, quality area size: 32, ref. point alignment area radius: 24, ref. point alignment search radius: 2