Hi Georg
This could be just noise, seeing, or guiding/tracking errors—remember that a 0.5” tracking rms, means you will have much larger deviations, some will be 2 or 3X larger than your measurement. Seeing changes can be very fast and guising is set up properly to ignore chasing seeing changes, but seeing can cause very large deviations that just are not picked up on the guiding time scale.
If it is noise, does the problem go away for the stack? Every sub has random variation, I find sometimes there is too much weight put on a single sub, which can potentially be noisy, or just subject to external influences like cloud, wind and vibrations. If the events are random, they will average out to a beautiful final, result. In the end, if it is just noise the FWHM of the final image will be near the average of all the FWHM for the subs, or probably a bit better. And the final star shapes should be nice and round. If that is the case, you are good to go.
However, a 60” luminance at that aperture seems excessive just to get enough stars for this analysis, so perhaps guiding/tracking is an issue. Find a field with lots of stars and take a relatively short exposure, even a 4 second or less exposure should give you hundreds of stars in a good rich field.
I don’t see it is reasonable your tilt is changing if you are just taking successive images. If your tilt changes over the night it might indicate something is moving in your image train, but that would be over a longer period of time—unless something is very loose. That has happend to me. Thus, I doubt that your tilt could actually be changing, and the SW is pretty robust, not sure how it could give you flaky results.
Hope this helps
Rick
