Conceptual
Problems Of Interactive Game Scores
As
I said earlier, music for listening is inherently non-interactive. The
compositional traditions of western (and most) music all require that
the final state of the composition is known and fixed. This requires
us to compose outside of many of these traditions, on new ground with
few guides. Much of the enjoyment we get from music comes when we know
the song well; Every nuance, every beat, every transition happens exactly
as we know it will, so in our heads we are able to "sing along",
and we feel ourselves merge with the music. (Best case scenario) We
know what’s going to happen, so why doesn't this become boring? In fact,
sometimes it does. To prevent that boredom, a good composer will throw
a variety of tonal changes and rhythmic cues that make the song pleasurable,
more so each time you hear it. This is exactly what begins to grate
on one's nerves while playing a game. Not only is the music not your
first priority and yet still vying for your attention, but even your
favorite songs will wear thin when played 12 times in a row. In a game,
with songs sometimes being limited to being 2 minutes in duration, you
may hear that song repeated 500 times in a long day of gaming. (And
again we scramble for the options page)
By
the same token, if one takes a good piece of game music, puts it on
a CD, and just sits and listens to it, it seems to be lacking. No vocals,
not much song structure you can latch onto.
And
the big problem is that most of the things that one can do to a track
to make it interactive are things that take it a step away from being
a well arranged song. For example, if you have to be constantly prepared
for the track to suddenly start playing a second track along with the
first, and that second track always starts at its beginning, then the
first track has to be always ready to be consonant with the beginning
of the second. That means you're limited to tonics (the main note (pitch)
that defines the song at any moment) which work with all the tonics
from the other track, and vise versa. This makes the composition linear
and lacking in variety, while a classic pop-radio song has a variety
of tonics that switch rapidly.
And
if for example you need a twitchy fast high-frequency flute line to
be heard every time a certain kind of enemy is nearby, you'll need it
to be there throughout your composition. The more things like this,
the more your composition will sound static, and the more potential
you create for conflicts with other lines. It's not an easy problem
to overcome.
Then
there is an issue with the "identity" of the song. We composers
are used to having a clear artistic vision of exactly what the song
should sound like. Composing an interactive piece requires that we imagine
our song sounding any number of different ways, and this is something
we're not used to.
So
how does one approach this situation without feeling set up to fail?
The first and most important prerequisite for making successful
interactive music is that the team commissioning the game has to be
ready to accept music that isn't like their favorite CD's or radio stations.
(Again, this does seem to be the trend so this problem is shrinking
each year.)
The
second is to free your mind from the fear of composing subdued, seemingly
uninteresting music. Trust me, when it's in the game, when it's part
of something, it will perform. I suggest buying all the old Brian Eno,
Harold Budd and Daniel Lanois 'ambient' (not 'ambient techno') albums
and listening to them on repeat for a month. That should cure you.
Third,
be sure to get all the exact spec on exactly what the proprietary interactive
driver the company you're composing for has can and cannot do. You won't
want uncertainty over whether what you're doing is even going to work
in the game adding to the complexity.
Finally,
think of your composition in an entirely new way: It's no longer a sculpture
of lines flowing from left to right. It's a basic skeleton with many
branching arms. Only some of those arms are heard at any moment. The
arms represent gamestates (character low on health, enemies nearby,
etc.) and have to work well with each other if there's a chance of them
playing at the same time. The branches have to shade the entire area
of the song because you don't know when they'll be needed.
Composing
Transitions
In
most cases, good interactive changes in music will not sound as if one
piece has stopped and another has begun. Most events in games, as in
life, are not that jarring. A 6-beats-per-minute tempo increase to respond
to the appearance of an enemy, adding a quiet high-hat percussion line
when the character runs, dropping some of the musical tracks by an octave
when the character goes underwater, adding reverb when the character
goes inside, these are all very effective. What can be overwhelming
for the player is large gestures for things that happen frequently,
like loud orchestra hits on every beat each time the character engages
in combat, or changing to what is essentially a completely different
song whenever an RPG character looks at his inventory.
Ask
yourself how long things are going to happen and compare them to your
music. For example, suppose you're composing for a game in which the
character can transform from a human into a wolf, and you'd like the
music to sound different. Try to know enough about the game to know
things like whether the player can make this transformation over and
over again in just a few seconds. If he can, don't make them all that
different. Not only will it make the music sound unpleasant, but it
may crash the audio driver. If this transition can only happen a few
times at certain controlled points, or if it's difficult to obtain the
ability to switch back right away, then they can be quite different,
as long as they sound like different versions of each other rather than
2 different songs.
In
this example, one could of course justify to oneself making the two
different music states into essentially 2 different, incompatible songs.
In my experience, the reason not to do this kind of thing is that the
more one does that, the more you unpleasantly grab the attention of
the player. It may wow the producers you have to please, and it may
sound cool when demo'd in the store, but after the 37th time, it will
sound embarrassing.
In
short, save the very big changes for very big things that only happen
a few times.
Also,
ask yourself if the time it takes for this variation to develop is too
long for the duration of the game event. For example, if you've decided
to drop only the melody by an octave whenever the character is underwater,
and that melody plays 12 measures at a time with 16-measure gaps between,
how likely is it that the player will go underwater and emerge again
without ever hearing the difference? Perhaps unmuting a constant low
sloshy tone the instant he hits the water is better?
If
you have a track that sounds very intense, which appears occasionally
to respond to some unusual intense state in the game (like walking on
fiery coals), try having gaps in the track. (Not long enough for the
player to go into and out of the situation without hearing the track)
The designers may expect that players won't spend too much time there,
but the players may do so anyway. Something that sounds intense enough
to represent walking on fire will also sound horrible if it goes on
for 48 seconds. Fade it in and out, or alternate it with silence or
something different.
________________________________________________________
Balancing
Transitions and Continuity
|
|
|
|
Features
