Quote -
I assume a JCM isn't a good solution because the morph would be linear, and the vertices need to move in a more complex manner?

The thing is, if she looks down and sideways at the same time, a different set of vertices would be affected. One could try to fake that with a whole cascade of morphs, but a magnet seems simpler. Still, the fine-tuning of the influence zone might be a problem.

Err, when I said parented to the eyeball, I meant wired up to follow the movements of the eyeball. It would need to influence the head actor (and only the head actor), though, so it'd probably have to be parented to that.

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