What about the results?
This is the rough data, no cleaned keyframes yet.
I didn’t get too much noisy positions on the 3D points as I had with webcams on earlier tests. I’m pretty happy with the results.
Here again I used flashy glass balls stuck on tracker paper bands.
On that test I used a facial performance captured separately, that explains why the real jaws don’t follow the low poly CG face motions.
Hairs are here because I’m a huge DBZ fan.