That doesn't make any sense whatsoever. You need to have a very specific idea of what you will have your characters do to be able to pose and position them in the environment, and for that you need to have a completed script for each scene. Case in point, how do you know what facial expression to give your character if you don't even have a dialogue? You drop in a joke, and the character needs to laugh, boom, new render. Sometimes I change half a sentence and need to re-render the picture it's tied to, because what's being said won't be in sync with what's shown.
See above. Definitely not 0% vs. 100%...