Graphics Reference
In-Depth Information
11.4.5 Appearance
In cases where the appearance of individuals is of interest (e.g., feature films), the objective is to pro-
duce a population with enough variability so the viewer is not distracted. If every individual in the
crowd had exactly the same appearance, it would noticeable to, and therefore distracting for, the
viewer. The same holds for detectable patterns in appearance. The objective here is to not have indi-
vidual features, or relationships among features, of the crowd noticeable to the casual viewer. A com-
mon strategy is to create mutually exclusive, yet compatible, sets with variations. For appearance, the
sets might be hair color, hair style, shirt color, shirt type, pants color, and pants type. By combinatorial
arithmetic, it is easy to see that even just 2 variations per set results in 64 different individual appear-
ances. This same approach can be taken for the movements of the individuals. For example, walking
behavior can be divided into the following sets: gait length, walking speed, arm swing distance, fre-
quency of head swivel, and frequency of head nods. Pseudorandomness can be an important part of
breaking noticeable patterns in the crowd. Even with the same gait, introducing slight phase shifts
in the walk can improve the visual impression of the crowd.
Because crowds can be viewed at a variety of distances, they lend themselves to geometric level-of-
detail modeling. The term impostor has been used to mean an alternate representation of a graphical
element that is of reduced complexity and is used in place of the original geometry when viewed from a
distance. Impostors can be used to reduce the display complexity of crowd members at large viewing
distances. Crowd members also lend themselves to level-of-detail behavioral modeling [ 26 ]. At large
viewing distances, the motion of the member can be simplified as well as the appearance. For example,
walking can be represented by a series of sprites (two-dimensional graphical objects overlaid on the
background) that have simple representations of leg and arm swing behavior. Because individual
behaviors are arbitrarily complex, there are opportunities to abstract out “typical crowd behavior”
in order to simplify modeling of the individuals (if the viewing conditions permit) when the behavior
of the individual member is only statistically relevant (e.g., [ 39 ] ). Actions can be selected on a statis-
tical basis and strung together to create the visual effect of a crowd.
11.6 Chapter summary
We are only beginning to understand the complexities of spatial reasoning, idiosyncratic behavior, and
generic behavior. There is no firm theoretical footing on which to stand when modeling personalities,
emotions, and memory like when modeling physical interactions of rigid objects. This represents both a
challenge and an opportunity for researchers and developers in computer animation.
References
[1] Blumberg J, Galyean T. Multi-Level Direction of Autonomous Creatures for Real-Time Virtual Environ-
ments. In: Cook R, editor. Computer Graphics. Proceedings of SIGGRAPH 95, Annual Conference Series.
Los Angeles, Calif: Addison-Wesley; August 1995. p. 47-54. ISBN 0-201-84776-0.
[2] Cassell J, Vilhjalmsson H, Bickmore T. BEAT: the Behavior Expression Animation Toolkit. In: SIGGRAPH.
[3] Cassell J, Pelachaud C, Badler N, Steedman M, Achorn B, Becket T, et al. ANIMATED CONVERSATION:
Rule-Based Generation of Facial Expression, Gesture & Spoken Intonation for Multiple Conversational
Agents. In: SIGGRAPH 94.
 
Search WWH ::




Custom Search