Hi, us again.... 
We are still in the middle of our character animation and "we want to have our cake and eat it, too".
Basically we want to have hi res models in our scene and real-time playback...
We've set up the following layers for the same character:
- Mr Box - just boxes for each joint (lowest res geometry possible)
- Mr Crap - low res model vaguely resembling the original character
- Mr Slick - hi res poly and subd model of the character.
Our self developed technique at the moment is that we copied (upstream inputs included) the models and distributed them over different layers. So when I activate Mr Slick (and hide Mr Crap) I get the same deformation as Mr Crap.
BUT before we put Mr Slick into scene Mr crap played back in almost real-time. Now since Mr Slick is included the playback of Mr Crap is jerky even though Slick is hidden (Not slow, but jumpy) which means that there must be some calculation going on in the background.
Mr Box is moving in real-time. But we want to optimize the playback for Mr Crap.
Any information regarding optimizing playback (pref. real-time) like caching, blocking nodes, proxy models, etc would be helpful.
Or is there a common technique designed for making proxy models designed for performance.
Cheers
