Lip Speak
I read in a review of "The Wild" a while back that the lip-synch was so good that it was "almost creepy". I haven't seen the film, but that remark strikes me as being rather odd. Surely if the dialogue animation was so amazing you wouldn't notice it, being so familiar with looking at talking faces all day long as we are. (Granted, anthropomorphised animals are something of a rarity in day-to-day life, but still...).
I reckon that animating to dialogue is one of those rare things in animation with a "right" and a "wrong" that the general public can pick up on in a snap. A good portion of our animal brains are devoted to picking up on facial characteristics so this definitely makes sense and a plausible reason to get it right. Contrast this to a character's weight being off, or something that snaps into place horribly. This would make animators everywhere wince, and possibly keen in misery, (in extreme cases). If it is noted at all by Joe Public however, chances are the viewer might only be able to define this in only the most nebulous of critiques - "it didn't look right", or, "something was off", and the like.
Check out my highly rubbish pic. An unsophisticated paradigm for animating a face to dialogue is to think of the audio as a simple series of individual phonemes that the mouth has to match in order to deliver the line. A common mistake I make is to always start off with too many mouth shapes. We go from a to b to c and keep hitting phonemes until the end, right? Sort of, but not really. While this may be true in the strictest sense, things are a little more organic than that. Doing it like 'crappy drawing A' you'd a mechanical jittery mouth that doesn't look like it's speaking. It's certainly easy to do, and to pull back from this takes experience and thoughtful planning...
Put your hand on your chin and say a line. You'd expect the jaw to go all over the place, but this is not the case. Do the same thing again but watch the corners of your mouth in a mirror. They are fluid and don't jump around crazily either!
A mouth is a bit lazy when it comes to making a sequence of shapes necessary to deliver speech. Like a "normal" walk, the body cunningly tries to get away with the least amount of effort required to do its task - in this case, speaking. We slur over mouth shapes, blend some together and even leave some out entirely. The point is to hit the most important phonemes and it will look right, and without the unnecessary chatter. So what are important phonemes? Vowels and sounds made from the front of the mouth are usually a good place to start, (P's, B's, M's W's, F's etc). As you get nearer to the throat (K, G, H) it's easier to gloss over these with the favouring of others. Choosing the right ones will depend on the situation, for what shape you hit and how hard you hit it is dependent on those surrounding it. See my very poor diagram for possibly no elucidation in this matter. It's kind of how my thinking about dialogue animation has changed since I started animating, (for the better!).
Phew. I think I started this post about 6 weeks ago but never got around to finishing it. It may be a load of twaddle, but it helps me by making me think about it in a new way.
I reckon that animating to dialogue is one of those rare things in animation with a "right" and a "wrong" that the general public can pick up on in a snap. A good portion of our animal brains are devoted to picking up on facial characteristics so this definitely makes sense and a plausible reason to get it right. Contrast this to a character's weight being off, or something that snaps into place horribly. This would make animators everywhere wince, and possibly keen in misery, (in extreme cases). If it is noted at all by Joe Public however, chances are the viewer might only be able to define this in only the most nebulous of critiques - "it didn't look right", or, "something was off", and the like.
Check out my highly rubbish pic. An unsophisticated paradigm for animating a face to dialogue is to think of the audio as a simple series of individual phonemes that the mouth has to match in order to deliver the line. A common mistake I make is to always start off with too many mouth shapes. We go from a to b to c and keep hitting phonemes until the end, right? Sort of, but not really. While this may be true in the strictest sense, things are a little more organic than that. Doing it like 'crappy drawing A' you'd a mechanical jittery mouth that doesn't look like it's speaking. It's certainly easy to do, and to pull back from this takes experience and thoughtful planning...
Put your hand on your chin and say a line. You'd expect the jaw to go all over the place, but this is not the case. Do the same thing again but watch the corners of your mouth in a mirror. They are fluid and don't jump around crazily either!
A mouth is a bit lazy when it comes to making a sequence of shapes necessary to deliver speech. Like a "normal" walk, the body cunningly tries to get away with the least amount of effort required to do its task - in this case, speaking. We slur over mouth shapes, blend some together and even leave some out entirely. The point is to hit the most important phonemes and it will look right, and without the unnecessary chatter. So what are important phonemes? Vowels and sounds made from the front of the mouth are usually a good place to start, (P's, B's, M's W's, F's etc). As you get nearer to the throat (K, G, H) it's easier to gloss over these with the favouring of others. Choosing the right ones will depend on the situation, for what shape you hit and how hard you hit it is dependent on those surrounding it. See my very poor diagram for possibly no elucidation in this matter. It's kind of how my thinking about dialogue animation has changed since I started animating, (for the better!).
Phew. I think I started this post about 6 weeks ago but never got around to finishing it. It may be a load of twaddle, but it helps me by making me think about it in a new way.



2 Comments:
Nice illustration. It gets the point across and that's the important part. There are so many subtle nuances that you miss if you just start plugging in blend shapes. "Twaddle?" I'll have to look that one up.
-Drew
By
Drew, at 11:46 AM
yeah, nice post kevan
By
cgAnimator, at 1:23 PM
Post a Comment
<< Home