I would like to add something here.
Don't use Force Global Quality unless you have transparancy problems that give you seams. Using local quality for each AP will generally give you a sharper image, especially when the seeing is poor, simply because one side of Jupiter is fuzzy doesn't mean the other side also is. Force Global Quality is just a trick to turn that local quality estimation off, which is useful under certain special conditions.
Noise robust 4 is 90% of the time just fine. For blue it might be better to go to 5 or 6, but that really depends on the seeing.
AS!2 is fast. This makes it relatively easy to test out these things. If you are not sure which settings to use, pick the best bunch of your recordings*, and play around with it a bit to determine what works best for each color. Go from large AP sizes to small, and then from large noise robust values to small, and carefully (!) compare the results.
Don't try to find rules for the quality graph, as I don't think there are any. (But if you found some, let me know
). One single really good or really bad frame will make a '50% rule' look completely silly.
*determined by using a single big alignment point with quality 4, which pretty much always gives decent results: good enough to see the difference between the recordings.