McCall.DEM is a collaborative work by Scot Gresham-Lancaster and Bill Thibault. It derives melodic, timbral, visual and rhythmic materials from a computer representation of terrain. The idea for the piece came from several sources.
One is Rich Gold's "Terrain Reader" program. Written in the late seventies, the program traversed a mythical terrain to produce sound. A "traveler"moved rapidly over the"terrain" stored in 256 bytes of the KIM-1 computers 1024 total bytes of memory. The elevation of the traveler at each instant was sent directly to a loudspeaker. The waveforms produced could be considered cross-sections of the terrain, cut along the traveler's path.
The idea of the Terrain Reader resurfaced during the planning of the Pleaides Project: a yet-unrealized plan to build a privately owned and operated radio telescope on the rim of the Grand Canyon. Composer David Cope enlisted Scot's help as Technical Advisor for the project. Part of the plan was to transmit music deep into space in hopes of contacting extra-terrestrials. In conversation about this aspect of the project Scot and I had the idea of transmitting music based on the terrain near the telescope site.
These ideas came together in June 1989, when Scot and I were considering what to do with a brand new NeXT computer made available to us at Calif. State Univ. Hayward (where we both work). The full features of the NeXT allow for a much more detailed terrain model and CD quality sound output far superior than Rich Gold was able to realize on a 6502 based single board computer.
We went to the United States Geological Survey in Menlo Park, Calif. to see about obtaining computer readable data for the Grand Canyon. Although available, the data would require several weeks to obtain, and would cost $60. We decided this was too much time and money for something that might turn out sounding worse than our expectations, so we settled for the "McCall Idaho Sampler. This contained data files of several types (land use, water, roads, etc.), including the type we were interested in, the Digital Elevation Model (DEM). This consists of a grid of elevations measured at 30 meter intervals over an area of approximately 100 square miles.
After massaging the data from the USGS into a convenient form, they occupy about half-million bytes of storage. One of the first tasks was to get an image based on the data, if only to see if it wasn't scrambled in processing. I modified a student project from the graduate course in Computer Graphics which I taught several months earlier to produce an image simulating the terrain as a metallic surface, lit from any angle. That data appeared correct. This program is used to produce a series of images during a simulated "day" by moving the light source along the path taken by the sun.
After seeing the data, we wanted to hear it. The first try was to simply run the rows of data directly to the sound hardware. The result was a noisy, somewhat periodic (pitched) squack. This didn't correspond to a traveler's path, however. To extract elevations from the data along a path, I decided to first break down each path into a series of straight lines. This would allow the use of a fast, well-known procedure for travelling a grid along a line between any two grid points: Bresenham's Algorithm. (The same procedure is used to rapidly draw lines on the grid of pixels in a computer display.)
After a couple of weeks it was possible to specify a traveler's path as a series of grid locations connected by straight lines. The traveller moves along the path at the sampling rate of a CD: 44.1 K grid locations per second. To get a tone of a specific pitch, a closed path of the appropriate length is traversed repeatedly: larger paths produce lower tones. (Glissandos can be produced by changing the length of the path after each circuit.) By choosing set lengths, a scale can be produced, and if those lengths are proportional a "just intoned" scale is produced.
Repeating the same path over and over produces a steady tone: not very interesting to listen to. If the path could maintain the same length, but change somehow over time, an interesting, tunable sound could be produced.
I tried an idea. A single line segment was the simplest path; the traveler moves from one end to the other and back to produce a single cycle of the wave form. If such a path rotated about its center, a changing tone based on the vicinity of the center could be produced. After solving a few technical difficulties, we had a rotating line segment which had a pitch that stayed constant. A "meta-traveler", located at the center of the spinning line segment, generated audio according to his surroundings. During this period, much of the implementation work shifted from the NeXT to the Amiga. The Amiga provided a simple programming environment, and furnished us with better physical access: we both owned our own Amigas. This simplified arrangements for the first performance of the piece by Scot at the New Music America festival, held in New York City in October of 1989.
That performance used slides photographed from laser printed NeXT images and topographic maps of the McCall area, two Amigas running the interactive code, playing Amiga local sound and a version of the traveler algorithm for Scot's 24 percussion robots (whackers), and a Mac using musical information produced on the NeXT to control external MIDI synthesizers. The Amiga program also supports MIDI synthesizers. Travelers can produce MIDI based on the terrain they travel over: for example, the pitch of notes can be congruent to altitude, high notes uphill and low notes down hill.
The travelers themselves can move on fixed paths or travel according to a BEHAVIOR. A few of the behaviors include: the "dry drunk": who stumbles about randomly, yet avoids falling into the lake; the "drunken Jesus" who moves over land and water; the "drunken sailor" who passes out at the helm of his speed boat, travelling in a straight line until hitting the shore, which wakes him up to shove off in a random direction and pass out again.
A similar behavior-driven aspect is a system supporting hundreds of independent particles that bounce around according to a simple physics in a faceted approximation to the terrain data. Collision velocities control parameters such as loudness of generated sounds. Current work in the area is focusing on modelling behavior of flying discs (frisbees TM) and wind driven blimps.
While working with these systems, a sort of mythology has evolved. Watching the particles bounce around the terrain in mile-high arcs led to calling them "bowling balls". Scot's wife, Kathryn, a poet, heard this and imagined the bowling balls being dropped from the mouth's of flying pelicans, and of the splashes when they hit the water near the drunken sailor's boat causing him to call out to the friends he imagines have fallen over board, and ....etc.
We have since performed this piece as a duo several times. Once in April of 1990 at CSUH in Hayward as a live interactive performance without whackers but with me playing live interactive computer graphics on a SGI Personal Iris and synthesizers. Scot performed with the same setup he had used in NY. Then later that year we did a four hour live performance in the studios of KPFA in Berkeley, CA. In that performance we performed both the McCall data and the newly acquired Grand Canyon east and west data in a sort of counterpoint.
In Nov. of 1992 we did a 4 day installation version of this piece that we called "Songlines.DEM" at the International Computer music Conference at San Jose State University in San Jose, California. That installation incorporated sites world wide as material for the algorithms. We, also, made several aspects of it interactive since it was more of an installation than a strict performance.
We envisage performance of this piece involving the interactions of the listeners with the terrain and the continued development of new and more dynamic performance travellers, and image generating techniques. It was obvious in working with this material that we had stumbled on an area rich with possibilities like discovering new land from the ether.
A one and half hour digital master was made from the material done during the radio broadcast. The digital editing was done with A A Bee Removal, Sam Ashley and Ben Azaram. If interested please let us know.
February 1993 Bill Thibault