Graphics Reference
In-Depth Information
t('o'). And the PI-2.duration attribute's value is determined as: t('i')
+ t('s'). Since none of the PI phases encapsulates a hold movement-
phase in this case, the PI.persistence attributes' values are, in both
cases, set at 0. The results of the temporal sync are two types of units
that store all the information necessary for the re-creation of the
generated coverbal gestures and their transformation into behavior
description. The Content unit stores the global temporal structure for
the corresponding conversational expressions, whereas the Phase unit
stores local information regarding overlaid shapes. In the following
section, we discuss the last deque that has to transform Content
units and corresponding Phase units into procedural descriptions of
synthetic behavior (behavior-plans), as required for virtual re-creation
of coverbal behavior (including hand/arm gestures, and lip-sync) on
synthetic ECAs.
5.4 Non-verbal generator deque
Most of the ECA-based animation engines re-create non-verbal behavior
based on procedural animation description mark-up languages.
Within the PLATTOS system, the
non-verbal generator deque
is used
to transform the coverbal gestures into a form understandable to the
animation engine. Therefore, the deque
transforms the HRG structure
into EVA-SCRIPT-based behavioral descriptions supporting both lip-
sync and coverbal gesture animation processes (Mlakar and Rojc, 2011).
Nevertheless, since the HRG structure stores very detailed information
on coverbal gestures, the transformation into other mark-up languages
is also simple and quite straightforward. Figure 10 demonstrates the
functionality of this deque. The deque needs Content units and Phase
units as input. The output of the deque is a behavioral script, written
in EVA-SCRIPT animation descriptive mark-up language. Further, in
order to re-create (animate) the conversational expressions as stored
in the HRG structure, those shape models determined by the Phase
units' attributes (
rhshape, rashape, lhshape
,
and
lashape
) are selected in
the gestural dictionary. These models already contain the movement
controllers' configurations (shape description), and must only be
temporally aligned, according to the temporal specifi cation, as specifi ed
by the Content (CU) and Phase (PI) units. Each CU unit is represented
by a '
bgesture
' tag (Figure 10). The obligatory attributes of the '
bgesture
'
tag are:
name
,
type
,
and
delay
. The '
name
' attribute is formed generically,
in this case, as:
“gestureCU-” + CU id
, as used in the HRG structure.
The '
name
' attribute is, in the context of animation, used only for easier
access to the selected procedure on the animation graph. Further, the
'
type
' attribute is generically set to '
arm_animation
', since hand and arm
Search WWH ::
Custom Search