Thanks for your response. I exposed for the shadow region. So I did change some initial exposure from when the moon is mostly dark to mostly white. However not enough to prevent highlight clipping. In other words the white is blown but the dark side has detail visible at all times. This is deliberate since one cannot expose to prevent clipping on both sunlit and dark sides given the high contrast. There is slight flicker since this is done by eye - but my problem is the centering not the flicker.
Does PIPP use highlights data to do the centering? It should be fairly simple to find a circle whether clipped or not and find its center I'd think? I obviously know only enough to be dangerous.
Ok. I think it is unlikely that PIPP will be able to align all the frames. You could try the surface stabilization method and select either a region of the dark side, if detail is always visible, or perhaps the section of the illuminated limb that does not go into the umbra, but I am doubtful either of these methods will work.
One thing you could try is Autostakkert without PIPP preprocessing. Turn up the brightness modifier such that the Moon is *always* blown out completely in every frame, and use planet stabilization. I have no idea if that will work. You can test it by playing through the resulting frames. If that works, you could then use the Expand function to chop the video in a large number of videos with a small number of frames, maybe 2 (I'm not sure it will stack just 1). Set the Noise Robust to 8, use Global, set the Reference Frame to be the same as the number of frames per chunk (e.g. 2), pave the entire frame (i.e. minimum brightness 0) with size 400 APs (you don't need Multiscale), RGB Align, stack to PNGs, stack 100% of frames, and let it rip. If it works, load the resulting stacks into PIPP and convert them to an AVI. Good luck.