Abstract
Animated characters abound in movies, games, and television. Their appeal depends not only on the roles portrayed by the characters, but also on their visual appearance and movement. There is a standard principle in computer animation that says that the visual complexity of the form of a character should match its motion and its environment (Lasseter, 2001). Previous investigations of character form and motion have, to a certain extent, supported this principle. The perceived quality of motion is known to depend on the degree of human-likeness of a character (Hodgins et al., 1998; Chaminade et al., 2007; Reitsma et al., 2008). We wanted to test this principle systematically by varying the appearance complexity of animated figures and their environments and measuring the effects of these manipulations on aspects of motion perception. We presented five animated actors (point light figure, stick figure, box figure, a human-like and a ‘superhero’-like figure) jumping across a clearly marked distance. The jumps occurred in two contexts of different complexity: a high-quality rendering of two cliffs and a more basic, low-quality background of a tiled floor. All characters were animated with one set of motions that were captured from a single human actor. Observers (N = 14) were shown the first half of the jump in context and asked to judge whether the character would clear the indicated distance on landing. Observers were able to predict the trajectory of the jump (83% accuracy, chance is 50%). However, there were no significant effects of manipulations of character form or environment. Our findings suggest that observers can extract information about motion in spite of large changes in appearance and context. Future work will examine the relationship between character form, motion and environment to determine how the measures (e.g., predicting trajectories vs. rating naturalness of motion) used to probe observers affect responses.
Disney Research, National Science Foundation.