The animation manifold
Characters that think, move and speak from one state.
Most AI characters are a voice with a face attached. The animation manifold joins them: where the learner stands, how the character feels and how she moves share one geometry.
One state
Thinking, moving and speaking aren't three systems stitched together.
Thinks
learning + personality
Where the learner stands and how the character feels, read from the learning manifold and the personality substrate.
Moves
expression
Face, body and gesture follow from that state, so a reaction looks like it means something.
Speaks
voice
Her voice carries the same state, and her mouth moves with every word.
One character, two places
The character in the episode is the one you talk to.
Live
Rendered in real time in the browser or a native app. The character listens, reacts and answers, and what she does next depends on where the learner is.
Studio
The same character rendered at series quality for episodes and songs. Children meet her on screen, then talk to her in the app.
Education products tend to have adaptive voice and no characters children care about. Studios have characters and nothing behind them that thinks. The animation manifold is where the two meet.
Built on the family
Two manifolds underneath.
The learning manifold, in production, says where someone is and what should come next. The personality substrate gives the character an inner state that holds. The animation manifold is how both become visible.
Status
In development.
Talking 3D characters already run in the browser and in native apps, and our studio pipeline produces episodes with them. The manifold that joins thinking to movement is what we're building now.