Quote - A quicker way might be to write a script which trims out all deltas which are smaller than a given threshold value.  Presumably the intentional deltas in the morphs are more pronounced than the background noise.

That should be useful, although there might still be problems if, say, all the morphs for the mouth have an extra shift that moves the mouth in the same direction.

Maybe one could combine both ideas and use the trimmed deltas to extrapolate the drift in the remaining vertices (assuming it is some systematic deformation and not just random noise). How good are you with least-square approximation? :laugh:

-- I'm not mad at you, just Westphalian.