Simulated annealing?

So, I’m thinking about mapping facial marker data, to a slider/pose based face rig, with something like simulated annealing. Think it could work well for this application (at least as I understand the general idea behind it).

Does anyone have experience with something like this?

Cheers,
Sune

I think you’re on the bleeding edge here. How you you be using an SA like technique? Would you have people mimic facial expressions from a known 3D rig, and then use SA to come up with the matches?