As a preamble, in case a buddying CG character TD was reading this article, please pay special attention the specific “Mouth shapes” term I am using here.
A Mouth shape IS NOT a blend shape. Blend shapes are single predefined shapes that can only go from 0 to 1 and as a result, it is impossible for the animators to make successful transitions between phonemes or make the lip sync look organic.
“Mouth shapes” are instead built by the animators by moving/rotating individual controls like the mouth corners, the lips, the lip rolls, the mouth Up Down, the “sneer muscle” control (Levator Labii Superioris) and their secondary and tertiary counterparts. (see Malcolm rig v2 demo for reference https://youtu.be/h3YKuj6qjAM?t=3325)
Now that we have cleared this up, let’s get back to the specific topic.
As I what I was watching the “behind the scenes” section of Aardman’s “Early Man” Bluray last week, I was particularly interested by a section showing the 12 mouth shapes used for lip sync at the Bristol studio.
Once I have a bit more time, I will share on this blog a little presentation I made at work for the animation team. In the meantime, let’s use my recent findings as an opportunity to talk about mouth-shapes in general.
Here are some screenshots taken from the Bluray followed by a recap.
The specific the shapes used are:
MBP / FV / DST / EE / AH /
OO / OH / CH / RR / KRN / TH / L
As CG animators, the first difference we can find is the lack of sad and happy shapes. Instead, Aardman animator seem to only require neutral shapes.
An other interesting thing is the lack of the UW shape, the lips are missing for the O shape so I could guess this is what the OH shape is or possibly what the RR shape is for?
The shapes look a lot like the ones used in CG except that in CG, it would be very easy to move the corners in and out to adapt the Submissive shapes (MBP / SDTK / L and FV) to the following or preceding Primary shape (O / U / E) in order to smooth out the transitions.
Stop motion features being, mostly, shot on twos, animators can probably get away with it as this is the charm of the medium. Having no control over the corners in CG would make the result very poppy and undesirable for most productions.
Just for comparison, I am reposting the Sony’s “Chester V”‘s mouthshapes and the Blair chart I featured a while back. (a link to the original article is posted below)
For “Cloudy with a chance of Meatballs 2”, Chester V had a lipsynch library of 16 shapes.
M / M2 / S / E / EH
I / L / A / AH / AO / OU / O / Th / U / U2
Personally, my library is closer to the Preston Blair chart as most shapes can easily be reconstructed from the following basic 10 shapes and a smaller library is easier to use and mamage.
A / E / O / U / NDTCDKNRSZ / W / MBP / L / FV / Th
I once came across a complex Kung Fu Panda lip sync library that would be useful for this article, let me know if you have a link somewhere.