Graphics Reference
In-Depth Information
movement in relation to the environment. In Allbeck and Badler's
model, this component is described using three dimensions: horizontal
(spreading/enclosing), vertical (rising/sinking) and sagittal dimension
(advancing/retreating). EMOTE acts on the animation as a filter. The
model adds expressivity to the final animation. It can also be used to
express some properties of the virtual agent or its emotional state. For
this purpose, the EMOTE parameters were mapped to the emotional
states (OCC model; Ortony et al., 1988) and personality traits (OCEAN
model; Goldberg, 1993).
Neff and Fiume (2002, 2003) propose a pose control model that
takes into account several features of nonverbal behavior such as the
timing of movement, the fluent transition between different poses and
its expressive qualities. For each body posture, different properties can
be defined like its tension, amplitude, or extent. The model allows a
human animator to vary, for example, how much space a character
occupies during a movement or to define whether the posture should
be relaxed or tensed. In more detail, Neff and Fiume (2002) propose a
model of tensed vs. relaxed nonverbal behaviors in a physically based
animation framework. This model is based on the observed relation
between expressive features of movement and forces of gravity and
momentum. The system allows an animator to control explicitly the
tension for each DOF in the character animation by taking into account
the gravity and external forces. Tense movements are short, greater
acceleration and without overshoot, while relaxed ones start slowly
and with a delay (necessary to overcome inertia) and finish with a
visible overshoot.
In a more recent work (Neff and Fiume, 2003), the same authors
focus on three different aspects of movement, i.e., timing, amplitude,
and spatial extent of gesture. First, they propose a sequential realization
of the movements where different joints are no longer animated at
the same time but they move in a sequence (i.e., some of the joints
follow the others) to make movement more fluid (natural). The
human animator can specify the offset and the type (forward or
reverse successions of joints) manually; then the system automatically
propagates the time shifts to all joints, which for this purpose are
organized in a hierarchical structure. They also model the amplitude
of the movements, i.e., they adjust the ranges over which a motion
occurs. The amplitude of the movements is modeled by multiplying the
distance of the joints for each pose (keyframe) of the animations. For
each inter-pose, two deltas are calculated: one measuring their distance
from the average to the end of the pose and the second measuring
the distance from the average to the end state of the previous pose,
then they are multiplied by the amplitude factor. The average values
Search WWH ::
Custom Search