Hi @zhenyu_tan,
To answer your first question, I believe the binning shouldn’t affect the particle locations – we record particle locations as fractional distance along the micrograph, so it should be independent of shape changes to the micrograph like during F-cropping and binning.
For the second – we have seen a similar issue before (Patch motion correction doesn't show any patch correction) where the trajectories were very jagged, and it appeared that the solution was that the dose parameter was incorrect (the dose-per-frame was input during Import Movies, rather than the total dose across all frames, which is what we expect). Does this perhaps explain the poor trajectories?
Best,
Michael