A phd candidate of the Utrecht University has developed an approach to natural gestures to animate by muscle and zenuwmodellen to use. He applied his models to two-legged, virtual beings, who themselves learned to walk and run.
Movement patterns in animated creatures are often calculated with physics formulas, making the creatures feel rather stilted and stiff to move, but the Dutch computer scientist Thomas Geijtenbeek figured out an alternative. He used natural muscle and zenuwmodellen for physically-simulated, two-legged characters, and determined the optimal location of the muscles by trial & error.
The main difference is according to Geijtenbeek in the incorporation of biological constraints. “For example, muscles have only a limited amount of power, depending on how long they are and how fast they contract. Nerves are relatively slow with the relaying of information, forcing people and animals, with some delay, react to sensory observations.”
The advantage of his method is that no complex programming is to the beings in a natural way to make them move. The modelling of muscles and nerves automatically displays. Also adjust the figures, their movement automatically adjust to the speed. They will run at higher speed, or are they jumping like a kangaroo. Another advantage is that interaction with objects or other characters is much more realistic.
Geijtenbeek phd in mid-december on his thesis Animating Virtual Characters using Physics-Based Simulation. He goes to his work continue to use it for research into muscular dystrophy, but also game developers would be the method to use, writes the University of Utrecht.