Graphics Reference
In-Depth Information
while maintaining a high level of motion quality, since each individual
gesture maintains the nuances of the original performance. A drawback
of this method is the large number of gesture performances needed for
each base pose or posture. For example, a different performance would
be needed for gestures when standing up with your hands relaxed at
your side, versus standing up with your hands on your hips, versus
standing up with only one hand resting on your hip, and so forth.
Posture changes, such as sitting down, crossing your legs and so forth,
would also require a new set of gestures for each posture. Note that
speech and facial movements can be simultaneously recorded with
the gestures when using the hub-and-spoke method, or synthesized
at a later time. Also note that using such a method limits the types of
performances that can be captured; the actor must return to the same
position that he started from, perhaps causing a lack of continuity
between utterances. However, this approach's strength is in its ability
to sequence gestures in any order or time as needed.
2.4 Blending gestures to achieve motion variation
While the best quality reproduction will be achieved by replaying
a performance exactly as it were captured, it is often impractical or
unfeasible to capture all the movements that would be replayed on a
virtual human. For example, a pointing gesture can take many forms:
accusatory, informational, subtle and so forth. Thus, variations in
gesture can be generated by multiple performances of an actor of those
gesture variations. This will result in a high-quality reproduction of
the gesture. This also means that additional motion-captured resources
will be needed to capture the additional gesture variation, and itself
will only be reusable for one additional synthesized performance
that matches the new variation. To remedy this problem, similar
gesture motions can be blended together in order to give a range
of variations between various example motions. For example, an
energetic beat gesture can be combined with a slow, deliberate beat
gesture to generate a gesture motion that appears halfway between
an energetic and a slow gesture. Likewise, directional gestures, such
as pointing gestures, can be generated with directional variations,
then recombined in order to form a pointing gesture in a direction
not explicitly captured, but rather synthesized as a combination of
two or more other pointing gestures. There are several limitations
to this approach; gestures that are to be blended together need to
have compatible characteristics in order for the blending to work
properly. In most cases, this means that the motions must have the
same number and type of phases. One motion cannot, for example,
Search WWH ::
Custom Search