Bzzzt! Nope! :banghead:
…
… 
I learned that lesson the hard way!!
Do it like they do in the movies:
[ol]
[li] Build your set and do your camera setups. Block out the action and preview it from all the cams. [] Create a “scene” for camera #1, linking to the root scene, select that camera, set the output file-name and length in frames. (Add a few extra frames for good measure.) [] Repeat for all scenes/cameras. (Caution: Be sure that each scene derives from the root scene, not the preceding one.) [] Render them all. (Use compositing aggressively to save time. See below.) [] Cut the film together. (e.g. using the built-in sequence editor.) [/ol]
[/li]
Most of the time, you go one step further, breaking the shots down by “RenderLayers” and “passes,” capturing all of this material into a MultiLayer (OpenEXR) format output file. Backgrounds for non-moving cameras, for example, need consist of only one frame; ditto non-moving objects in such shots.
A compositing pipeline is used to actually build the finished strips, and adjustments can be made on-the-fly. With this approach, most if not all of the image changes you’ll need to make can be made without re-rendering.
So, the process of generating the film is nothing at all like the finished film will (appear to) be when it’s stitched together. And for that matter, the various channels of material that come out of the render, until they’re composited, don’t look too much like the finished shots they’ll be used to produce.
Forevermore put out of your mind the notion that “the finished scene pops out of the renderer, and if something’s wrong I’ve no choice but to render it again.”
(Like I said, I learned that lesson the :buttrock: hard :buttrock: way.)
Please take this in the friendly spirit intended! 