Graphics Reference
In-Depth Information
The use of motion capture extends beyond capturing and recording figurative
movement: it also incorporates the capture of facial expressions. Motion-capture
techniques and processes used in conjunction with an actor as a subject have
created some remarkable performances. They played a major role in creating
the performance of Gollum in the
Lord of the Rings
films. This was achieved by a
mixture of the movements of performer Andy Serkis, who gave life to Gollum,
and the CG animators, who used more traditional keyframing techniques. The
movements are captured as data, and while they are conformed around the
general physiognomy of the performer or subject, they do not include all the
performer's physical details. Therefore the movements of one subject may be
easily mapped onto a character of different design. The technique allows for a
speedy and relatively inexpensive way of creating naturalistic action on which
the animators can then build. This efficiency has been a boon to the features
and games industry, which uses these systems to great effect.
Camera moves may also be recorded using these techniques so that the
performance, captured digitally, can be mapped onto previously recorded
live-action camera moves. Match moving allows for camera tracking of digital
camera moves and the live-action moves and for accurate compositing of the
different elements to integrate them seamlessly into the same scene. Camera
tracking and match-moving techniques are not only responsible for some of
the more spectacular shots in feature films; they are also starting to appear
regularly in other forms of filmmaking, including wildlife documentaries.
There are a number of systems available for motion capture. They fall into two
distinct areas: optical and nonoptical systems.
Optical systems
operate by tracking markers placed on the subject that are
detected by an array of electromagnetic sensors positioned in such a way as
to create a performance area. Within this area the movement is triangulated
by a given number of sensors. This number may vary; the greater is the
number of sensors, the more detail that can be gathered on the subject. This
raw data then needs to undergo a level of processing before it is applied to a
computer model. Animators may then work with this automated movement
to refine or add additional movements.
Nonoptical systems
fall into several categories. Some systems use a series of
gyroscopes to make the necessary measurements of movement, but these
inertial systems currently lack the accuracy of other systems. However, they
are much cheaper systems than optical systems.
The mechanical motion system gathers motion data directly from sensors
placed at specific points on a data-capture suit. This works rather like an
exoskeleton because the sensors form a rigid structure comprising a number
of linked, jointed rods that correspond to the subject's physiognomy. Sensors
placed at the joints then detect the relative movement of the performer's
arms, legs, and so on. The nonoptical mechanical systems clearly have their
applications, but they are capable of achieving more limited data-capture
volumes than the electromagnetic optical systems.
