Graphics Reference
In-Depth Information
are applied to poses from the second to
n
- 1 in a sequence. A similar
mechanism is used to calculate the spatial extent, i.e., the space where
the action is realized.
Hartmann et al. (2005) define and implement a set of parameters
that allow one to alter the way in which an agent expresses its actual
communicative intention. This model is based on perceptual studies
conducted by Wallbott and Scherer (1986), Wallbott (1998), and
Gallaher (1992). These works define a large number of dimensions that
characterize gesture expressivity. Hartmann et al. (2005) implement
six of these dimensions (see Section 3 for more details). Three of
them, namely spatial extent, temporal extent and power, act on the
parameters defining the gestures and the facial expressions. They
modify respectively the amplitude of a signal (which corresponds to
the physical displacement of a facial feature or the wrist position), the
movement duration (linked to the execution velocity of the movement),
and the dynamic properties of movement (namely acceleration).
Another dimension, fluidity, modifies the movement's trajectory as
well as works over several behaviors of a given modality. In the latter
case, it specifies the degree of fluidity between consecutive behaviors.
The last two dimensions, overall activity and repetitivity, refer to the
quantity of signals and to their repetition.
Szczuko et al. (2009) propose a fuzzy controller that can be used
to modify a key-frame-based animation by applying two expressive
features: fluidity and level of exaggeration. In this approach an
animation is modified by adding some additional gesture phases,
namely preparation, overshoot, and phase movement hold. The
preparation is a slight movement in a direction opposite to the main
direction, while the overshoot is an additional movement of the
last bone in the chain that overpasses the target position and goes
back. According to the authors, adding these phases to an animation
modifies its expressive quality without changing the communicative
meaning of the displayed behavior. In the paper, the authors propose
an approach that allows a human animator to specify explicitly
the fluidity of animation as well as its expressiveness through a
set of discrete values: {fluid, middle, abrupt} and {natural, middle,
exaggerated}. This approach is based on fuzzy rules defined in a
perceptual study. According to the given values of expressiveness
and fluidity, the fuzzy rules control the duration and the amplitude
of additional phases.
A similar approach, proposed in Kostek and Szczuko (2006), can
be used to generate emotionally characterized body animations. In
this approach, the authors first manually create a set of animations
of certain gestures. Second, the emotional content and intensity of
Search WWH ::
Custom Search