Best thing I’ve found so far is to decide, first of all, approximately how long I want the sequence to run, and what “marks” (significant poses) I want the character to “hit” along the way. These are the initial keyframes. And, the character that is now being animated consists of cubes, cylinders, pyramids, and other simple things.
In the background, I put various planes and simple blocks to act as backdrops.
Everything is actually to scale. Say that “1 BU = 1 meter.” Okay, even the blocks and pyramids (and backdrops) are appropriately scaled. Ditto the backdrops. There is zero detail here, but perfect accuracy. (That’s critical.)
Using the OpenGL Preview feature exclusively now, I also consider where I want to put the camera … and, I put several cameras down, at a variety of positions and lens settings. Each camera is named. Camera positions (and those being considered) are marked with named empties. (Like the little squares of gaffer-tape on a real set.) Soon, the floor is littered with cameras. (Fortunately, this being CG, none of them are “in the way” and there’s not a cable in sight.)
Then, I “shoot film” from all those cameras, with labels including frame#, camera name, scene, file. I put it into the video editor of my choice, and try to do a rough cut. If I think of another camera angle I’d like to use, I add another camera, keeping all the old ones, shoot the strip from that camera, and continue. I use “scenes” to capture the current camera-selection and the name of the output file, so each one is distinct. (At this point, “a movie file” is just fine.)
What you ordinarily think of as “the timing of the animation” is influenced, not only by the movements of the character, as dictated by the choreographer and stage director, but also(!) by the subsequent editing decisions made in the editing room. Really, those editing decisions are where “the timing and pace of the show” are made.
This is where you use the viewer’s powers of imagination to save time and create drama:
You show the pitch … CUT TO the umpire’s point of view as … CUT TO the batter tenses … CUT TO (ECU) the bat connects “crack!” … CUT TO the ball in the air … CUT TO the batter starts to run … CUT TO he leaps against the wall … CUT TO (ECU) the ball bobbles at the top of his mitt and flips over the wall … CUT TO it plops gently into someone else’s glove … CUT TO (ECU) the face of an excited little boy … CUT TO the mixed emotions in the face of the father, whose little boy just caught the ball, but whose team (upon whom he had mortgaged his house to bet heavily) just lost the game, and who will now lose his house.
(Add a few anticipatory shots of the father and/or the little boy, season to taste.)
Basically, a direct adaptation of the “real” film-shooting and film-editing process.
OpenGL Preview can churn out “film” at near real-time speed, making it very easy to explore possibilities because footage is no longer expensive to produce. And yet, you can (and should …) get very precise about it, because: these “preview” renders will match exactly a full-on render of the same thing. OpenGL can do a very good job of lighting and so-on. (In fact, OpenGL outputs can be part of the final show! If not a complete frame, then certainly a compositing layer.)
If you “edit first, then shoot,” you’ll find yourself winding up with a completely different notion of “what film is needed and what isn’t.” So, you shoot film freely, “in case” it “makes the cut.”
When you are finally staring at a little pyramid-man putting on a cool show, you now look at your edit list, and that is what, you now know, you need to model, texture, light, and render. (Render a second or so to either side, if you can, to give you some editing-discretion in the “final final cut.”)
