The three roundtable sessions were attended
by a wide variety of programmers, sound designers, producers as well
as representatives from various sound chip companies with direct interest
in 3D audio.
Although the roundtable was not focused on DirectX per se, one of the
first topics to come up in all sessions was Microsoft's DirectSound3D
(DS3D), which shipped with DirectX 3 in September 1996. Most of the
developers and producers had not yet had a chance to use DS3D or were
just starting to look at it. Comments on DS3D from those who had ranged
from "finally a 3D sound API" to "yea, but it doesn't have X or Y" to
"nice, but it's so slow." Fortunately, in two of the three sessions
there was at least one programmer with significant DS3D experience to
share their knowledge. Their basic suggestion: "Yes, DS3D has some significant
limitations. Work within them, and you can do some cool stuff." For
example, CPU usage of DS3D was brought up as a major concern (too much
CPU used for a 3D sound). That can be solved by letting the player determine
how much CPU the sounds should get. In the programmers' game, this was
accomplished with a "sound realism slider" that ranged from 1 to 10,
where 1 was mono and 10 used DS3D. He also suggested choosing judiciously
which sounds should be in 3D and which can get by being normal stereo
or even mono. Everybody in the sessions thought that was a very good
idea, although there was some grumbling that that meant a lot of work
for the programmer. Discussion also turned to the DS3D API itself. Top
on the list of suggestions was a method for doing room modeling or echo.
The example was given of a car driving through a tunnel and back into
the open. The car should get 'echo'y' when in the tunnel, but be 'dry'
when it comes out. At this point the chairman of the ia-sig 3DWG (Interactive
Audio-Special Interest Group 3D Work Group) said that the ia-sig is
working closely with Microsoft on future versions of DirectSound and
is looking for just this type of input, and encouraged suggestions (the
ia-sig can be contacted at mma@midi.org).
DS3D discussions also covered the issue of hardware acceleration of
3D audio. The analogy was made with graphics cards: "I want my game
to run better with 3D audio acceleration, but I don't want the game
to depend on it." was one producers cry. At the same time, a programmer
raised a concern of the amount of time involved in supporting two entirely
different sound systems, one for non-accelerated systems (using DS stereo
API) and one for accelerated systems (using DS3D API).
Another topic addressed was the current limitations of 3D audio technology
and how those limitations could be addressed by content creators. An
example was brought up about how to deal with the player who has very
cheap speakers, since the 3D audio effect may not be very noticeable.
One reaction to that was "If they've got cheap speakers, too bad they
can't hear the 3D…..if they want a better game experience, they've
got to make at least a minimal investment." Another developer however,
pointed out that that was a limiting concern. Since he couldn't depend
on proper delivery of 3D audio for all his players, he would be reluctant
to make it an essential game-play element. Another avid game player
and sound designer brought up yet another point: even if you've got
good speakers, you might not be right between the speakers, so you lose
some of the 3D effect. One attendee suggested that players should use
headphones to get the best 3D effect, but it was pointed out that you
can't force everyone to use headphones, especially on systems with built-in
speakers and hard-to-locate headphone jacks.
That led into the topic of how to best utilize 3D audio. Obvious examples
were given, such as assigning 3D coordinates to sounds that match 3D
graphical objects, such as helicopters, enemies, etc. as an aid to the
player. One programmer said they thought 3D audio could be used to "stretch
the screen" so the gameplay area became more than just the monitor,
but all of 3D space around the player. Other uses for 3D audio that
were brought up by a sound designer was using 3D audio to create environments,
such as jungles, ocean scenes or sports stadiums. It was also pointed
out that environments can be created entirely in a studio and given
to the programmer as a normal stereo wave file, so the 3D effect could
be heard without needing a special 3D audio card. Also mentioned in
all three sessions was how important it was for audio to be introduced
early in the game design process. Although that is an on-going theme
in the game-audio community, it was thought to be especially important
by the group when using 3D audio. By avoiding the "put all the sounds
into the game in the last 4 weeks of the project" problem, sounds can
and must be more carefully integrated into the game and game play.
In summary, all of the participants were very excited about the prospects
for 3D audio and several were planning on using some type of 3D audio
in their upcoming projects. Although there are limitations and obstacles
to overcome, the overall mood was very optimistic.
