Information Technology Reference
In-Depth Information
objects; and the buzzing from vibrating objects
in light contact with each other. In addition, the
project can produce the effect of surface damping
(an object becomes less resonant due to being in
contact with another object) and also the effect of
a change in sound due to a deformable object be-
ing forced out of shape. To a first approximation,
the effect of distributed nonlinearities described
earlier can be simulated, producing interesting
results like pitch glide. Further, as yet unreleased,
functionality has demonstrated the effect of dif-
fuse resonance which is when an object's modes
“become very numerous and merge into a diffuse
continuum” (Menzies, 2007, p. 3). More recent
demonstrations of Phya have introduced loose
particle surfaces, for example, gravel, achieved
through a PhISEM like approach, as well as sur-
faces covered in leaves, plastic packaging, and
shallow water. Bear in mind the author does not
claim these effects are created through accurate
modelling of the physical phenomena that cause
them but, instead, techniques have been developed
by combining an understanding of these phenom-
ena with a creative approach to sound synthesis
while reserving some aspects to be controlled by
a sound designer. In a later publication, Menzies
refers to this as “the development of semi-physical
perceptual models that provide some freedom for
the sound designer to more easily mould a virtual
world” (Menzies, 2008, p.1)
The information required to create these sounds
is unlikely to be available directly from a physics
engine and therefore a collision update layer of
Phya must deduce it from what is available. Often,
the physics engine will not provide enough detail,
for example when multiple impacts are involved,
and so the collision update layer must generate
extra information from what is known in a deter-
ministic or stochastic way. The collision update
process is reported as one of the more tricky
problems the project has overcome, with the issue
of monitoring continuous contact said to be par-
ticularly awkward as most physics engines do not
use persistent contacts but instead simply report
if two objects are touching in a given frame.
Before concluding, Menzies, who himself has
experience in both the academic and industrial
sides of computer game development, gives his
own “tentative explanations” as to why physical
modelling hasn't been embraced outside of re-
search projects despite its potential value. Firstly,
as is widely known among sound designers and
audio programmers in the games industry, audio
is usually considered less important than graph-
ics, meaning that less resources are allocated to
it both in terms of development and hardware.
Therefore “audio programming is often carried
out by a non-specialist programmer” and “there is
often a natural resistance to acknowledge that out
of house technologies could be valuable if they
can not readily be reproduced in house” (Menzies,
2007, p. 5). Considering the difficulties associated
with correctly harnessing the information available
from a physics engine, it is perhaps understandable
that developers may not wish to risk spending
resources on a concept that has yet to be tested in
a commercial sense. In addition, Menzies contends
that “published research often focuses on a level
of audio modeling detail that goes well beyond
that required in a simulation” (Menzies, 2007, p.
5) implying that current research is not entirely
relevant to an industry where a level of creative
control is desirable, algorithmic efficiency is vital,
and absolute authenticity is not.
In 2007, Raghuvanshi and Lin, now collaborat-
ing with Christian Lauterbach, Anish Chandak,
and Dinesh Manocha all from the University
of North Carolina, published an article on their
continued research. Real-Time Sound Synthesis
and Propagation for Games (Raghuvanshi et al.,
2007) reviews their previously described work
(Raghuvanshi & Lin, 2006) before presenting
sound propagation functionality that has since been
added. Their approach is an adaptation of beam
tracing (Funkhouser et al., 1998) and it is, as they
claim, well suited to interactive applications. Their
results indicate that, in complex, dynamic scenes
Search WWH ::




Custom Search