Hi,
I tried running WarpTools fs_motion_and_ctf versions 34 and 36 with exactly the same command and pointing at the same eer input. It outputs tilt averages that look identical but have a different pixel value range on absolute scale. Here is the info from the the same tilt average mrc headers:

V34:
Minimum density ....................... -71.314
Maximum density ....................... 1201.4
V36:
Minimum density ....................... -5.0651
Maximum density ....................... 68.827

Further protein average from v36 is much weaker compared to v34. Relion 3d refine converges well with particles extracted from tilts processed in v34 and falls apart when extracting from tilts processed in v36.