• face - (Poggio research)
• body - (a la poser in 2D)
• control points at important joints on the body
• stick figure representation
• rotator controls to adjust turn of body
• onion skin templates, with lights on contols to guide user
• 2D/2.5D? more complex with higher dimensionality
• build up vocabulary of gestural poses in scrapbook
• infer between different emotional states (via controls).
  If emotion is described as pertubations from norm, then by combining pertubations, perhaps we can create more complex gestures.
• on screen motion between body splines to express action not pose on screen
• fade out of emotes (roy...) to make it timely
• small multiple for face emotion change (also body?). only if pertubations from norm.

how do you express "hug" with avatars?