Speech

Neva Again by Vordul Mega



This took about 5-7 hours and four attempts. I tried doing it straight ahead by imitating the mouth movements and drawing random mouth shapes, recording myself and copying it frame by frame, looking at The Animator's Survival Kit for reference (I also watched the Making of A Nightmare Before Christmas), and finally tried to spell out the words being said frame by frame.

Here's what I learned:

Dialogue for animation is dominated by vowels. If you wanted to simplify it you could use six or eight basic mouth shapes: (1)A, (2)E/S, (3)I, (4)O, (5)U/Sh, (6)M, (7)P/B (8)F (i did that on the fly so i might have missed lots), some in-between mouths, and, of course, some random ones for giggles.

Programs that automatically lip-sync animation like Flash and Toon Boom use a system like this where they try to detect these sounds and use one of these images for it. This, however, almost always looks horrible, unless there is a bit of randomness in there (or you can design your mouth system well). The article I mentioned says Hanna Barbara used a 6 mouth system, which makes a lot of sense when animating a hundred million cartoons, but it does dimish "belivability" if the voice makes the mouth look dubbed.

I always used some random mouths snear or grins, and noticed differences in accents. Some people have a "z" in their words: pops vs. momz, for instance, which makes z an exaggerated s.

Once you've got your mouths, you plan out your dialogue on a exposeure sheet (X-Sheet/Dope sheet), which lists frame by frame how many drawing you will need (up to 100 on a sheet). You will have to listen to your sounds and try to map out which frames they are being said on so you can position your mouths accordingly. Use either the words themselves broken up into syllabals, or the sounds: "popz" would be P-A-M/N-Z, a P mouth shape, an A (which is the key), a closed mouth (M) and a Z. If you wanted to skimp, you could get away with jsut doing a A cushioned by two M's on either side (wiggaddy-wack).

I also noticed a delay effect where the animation works better if you position your mouths a little bit ahead of the moment the vowel is sounded. There's also lots of anticipating and reaction going on there. This song is a perfect example becuase you anticipate keys with the beat. Every time the base hits, your brain makes that the key frame which hits and slowly fades as the next base hits. I almost want the animation to do this two-step thing, going back and forth in a cosine curve. In the Survival Guide, Williams talks about the masters having a built in timing in all their animations with the best example being Road Runner. Everything is beautifully timed meaning there is a frame by frame rhythm and pattern throughout the entire animation. Musicians pick this up and can either run with the built in tempo, or go against the tempo to mess with anticipating and reaction. They're swiss watches, man, or OSX's. Most people will never know how elegantly the code is written. Anyway, so just like every word has a emphasis, so does every sentence. Dialogue, then, has a pattern depending on who's saying it.

All and all, I have a a pet peeve about mouths cuz of anime vs. cartoons, where anime is infamous for skimping on mouth movements. I hate it. Plus, i like angry expressive characters.