COTA 750: Expressive Motion
Animation of a quadruped, Canis familiaris (breed-Greyhound), trotting
Since April 2002, I've been motion capturing dogs in order to build a learning module for veterinary medical students to learn about patterns of movement associated with various types of lameness. I have had trouble with binding models to the motion capture skeletons within Maya. The geometry collapses in the chest and the bones flip unnaturally (90* to 180* rotations), mostly at the heads of the femurs and shoulders.
I'd would like to write a MEL script to convert my forward kinematic motion capture data into inverse kinematics. I'm hoping that once I do this, I will be able to fix "flipping" of bones in my motion capture data just as we did in Assignment 4. I could then try binding a dog model I found on the web to see if the bones still flip.
Alternatively, I can try to get rid of the flipping by creating an offset skeleton and constraining locators on it to locators on the motion capture skeleton. The offset skeleton, without any keys set on it, should then follow the motion of the motion capture skeleton.
If all fails with using the motion capture data, I can always create the skeleton and motion within Maya by keyframing by hand.
Images of binding problems
For reference: illustrations of leg movements during trotting
The pattern of movement associated with a healthy trot in a borzoi would be right front down, left back down, followed by right back down, and left front down. The illustration above was drawn by Robert W. Cole and was taken from Dog Locomotion and Gait Analysis by Curtis M. Brown, 1986.