Here's a tip I've seen from many in the plastic modelling field: paint what you see; if you can't see it, then paint what is likely there; and if that fails, then paint what you like and let your critics prove you wrong.