Hi,
Would it be possible to calculate the junk removal masks prior to motion correction (just on the averaged stacks)?
This could be helpful for two reasons - one, to discard the really bad micrographs prior to motion correction (and thereby save disk usage and compute) and two, potentially for future usage of these masks to exclude these regions from interpolation during patch based motion correction or patch CTF estimation.
For example, surface contaminants will not necessarily show the same beam induced motion as the underlying ice, and gold and crystalline ice contaminants could throw off local patch-based CTF estimation.
Cheers
Oli