Motion capture is the process of sampling the posture
and location information of a subject over time. The subject is usually a
person, an animal or a machine. In case the subject is a person or animal, it is
sometimes referred to as an 'actor'. Especially in the entertainment industry,
motion capture is also frequently abbreviated as 'mocap'.
The technical goal of motion capture is to get the motion
data of certain points of interest on the subject, so that either some
parameters of the motion (e.g., speed, angle, distance, etc.) can be calculated
or the data can be used to control or drive something else. In case parameters
of the motion are calculated, the application may be motion analysis, sports
analysis, biomechanics, biodynamics, etc. In case the data is used to drive a
computer generated (CG) character or a scenery to mimic the motion, it is
referred to as animation or visual special effects (VFX), etc. In case the data
is used to control a machine, the application may be tele-surgery,
tele-robotics, motion feedback control, etc. In case the data is used to control
some displays or something else, the application may be virtual reality,
interactive games, virtual training, virtual rehabilitation, motion directed
music, etc.
|
MARKERLESS MOTION CAPTURE
|
The earliest form of motion capture was done with cameras
taking a sequence of images (pictures) of the subject. Then the images are
analyzed by people to figure out the 2D motions of certain points of interest of
the subject. This technology is still being used today, except it has been
advanced to become 3D in nature by using multiple cameras to simultaneously take
multiple images of the subject from different directions, and software
techniques have been developed to analyze the images by computer instead of by
people.
The most difficult task of this motion capture approach is
recognition of the points of interest, which seems easy in concept for human
being but is highly difficult and time-consuming to do by computer software. The
achievable accuracy of the recognition is also insufficient for many
applications.
The difficulty of accurately recognizing the points of
interest from marker-less images led to development of a new class of motion
capture technologies which mark the points of interest explicitly with either
sensors or markers. The electro-magnetic, mechanical, gyro, accelerometer and
optical fibre based technologies mark the points of interest with sensors, while
the optical technologies mark the points of interest with markers. The markers
of the optical technologies are in turn distinguished as either 'passive' or
'active'. Passive markers do not generate light by themselves (e.g., reflective
balls or checker-cross patches), while active markers do (e.g.,
LEDs).
Motion capture systems are also classified as
'self-contained' or not. Normally the mechanical, gyro, accelerometer and
optical fibre systems are 'self-contained', meaning that they do not contain any
other part which is separate from the capture subject. The electromagnetic,
optical and some gyro-accelerometer systems are not self-contained and consist
of parts which are separate from the subject which must be fixed relative to the
ground.
|
DATA REFERENCE & CALIBRATION
|
The motion data captured by self-contained systems are
normally referenced to the initial posture of the capture subject. This means
that they can normally provide only relative posture information of the subject
body. The absolute location of the subject, if provided, is derived by
accumulating the previously sensed relative location movement information.
Unfortunately the accumulation process ('integration') also accumulates any
errors in the previous information, hence any absolute information provided by a
self-contained system is prone to large errors which are referred to as 'drift'.
To fix the problem some of these systems supplement their basic structures with
additional sensing parts separate from the subject and lost their self-contained
status.
The motion data captured by non-self-contained systems
are referenced to the parts fixed relative to the ground. Unless the fixed parts
moved, the absolute positions and location information of the subject will not
drift. Should the fixed parts experienced any movement though (e.g., due to
temperature variation or touching), a 'system calibration' process will need to
be done in order to re-establish the data reference information (the 'coordinate
reference frame'). This process involves collecting a relatively large amount of
data from the capture space. Depending on the system being used, the process can
be quite tedious, the user may be required to understand how it works, and it
may need to be done frequently especially if the fixed parts are installed over
a large and/or non-rigid frame.
WHAT TO CONSIDER BEFORE
BUYING
|
Motion capture systems have been developed with at
least seven distinct technologies as mentioned above. Depending on the
application, deciding which system to use may be a brain teaser. Following are
some of the questions one should consider before making a decision:
- How large capture space do you need? - How much
larger space do you have for setting up the system? - Do you require high data accuracy or not? - How complex
motions do you need to capture? - Do you need to capture motions from special
angles? - Do you need to use the captured data
immediately (for 'real-time' applications)? - In case of real-time
application, can it sustain occasional system crashes due to the technology
used? - Do you mind having to learn well and do the system calibration
manually? - How much calibration work can you withstand? - Will the system
users leave regularly and new users need to re-learn the system all over
again? - How much can you withstand having to post-process the data before
using it? - Do you need the system to be portable and easily installed for
use in different locations?
Motion capture system prices and features differ
significantly. A wrong system can cost you much more during use than the extra
price of another system. Many systems are being turned into junks because nobody
has the time or interest or know-how to learn to calibrate them, to post-process
their data, to fix their crash problems, or to re-configure them for capturing
special motions. Therefore unless a large internal support resource will be
available, the 'ease-of-use to yourself' should be placed high on your list of
factors to consider before purchasing a system. |