"I went to school at theUniversity of Illinois and was very fortunate to find the experimental music studio in Champagne/Urbana. They were looking for somebody to wire things together and I got the job. The University of Illinois was a very happening place in the late sixties because of Lejaren Hiller, Herbert Brun and a technical guy named James Beauchamp who was actually an electrical engineer. That was in 1968. We had one of the first Moog synthesizers and we had built something called the Harmonic Tone Generator. Sal Martirano was one of the most progressive and daring of the music faculty and was very kind to all of the students who showed any interest at all. He'd invite them into his home and we'd sit around and have discussions and play music. I admired him immensely because he took it upon himself in his middle age to learu electronics, circuit theory, and digital logic in order to progress his art.
DIRECT VIDEO SYNTHESIZER
(ZERO AND ONE)
The first Beck video synthesizer was later called Direct Video Zero. Direct Video #0 (DV #0) was an expansion of Beck's Illinois experiments, consisting of a modified color television set, with modulation sources driving the color CRT's red, green and blue electron guns. Colors were formed from oscillators and audio signals combined with external analog mixers. The modulation sources were pulled from a Buchla Electronic Music synthesizer to visualize sound. These color images were named "Direct Video" by Brice Howard, director of NCET.
The difficulty of using audio that "sounds good" to form an image that "looks good" was problematic in DV #0. The most interesting images were found from sound sources which were harmonically related to the vertical field rate (60 HZ) and/or the horizontal rate (15,750 KHZ), frequencies not common to audio synthesizers. The search for dedicated sources of video patterns, and a grant from the National Endowment for the Arts in 1971 evolved into the Direct Video Instrument One (DV #1).
The central element of DV #1 to generate the "direct video" image was called by Beck a "voltage to position converter". The converter was loosely based upon a "wipe generator" of a conventional video switcher. The wipe generator consists of a horizontal and vertical locked ramp generator locked to the horizontal and vertical sync. The ramps are compared against "wipe voltages" from knobs to determine the size and position of a switching signal that appears to "wipe" one image over another. The wipe circuitry was modified, replacing the knobs with voltage control of its operation. An input voltage changes the size and/or position of the waveforms triggered by the comparison point along the horizontal or vertical axis. DV #1 modularized this converter, then added an edge extracting "Outliner" that was wired to binary logic gates. The combined signals were patched into multiple color voltages summed together to feed an RGB to NTSC Color Encoder. The use of the NTSC encoder replaced "driving the guns" of the CRT in DV#0, and enabled the results to be recorded on video tape. DV #1 was constructed in a rack mount chassis with two rows of modules and patch cables formed from 1/8" mini-phono plug cables. The modules include:
1) Two dual axis joystick controls
2) A Horizontal and Vertical Ramp generator
3) A H or V phase-locked voltage controlled oscillator generating a triangle and square wave output. Non-linear waveshaping was later added.
4) Eight Voltage to Position Converters - switch selected on H or V, to generate rectangular pulses. These pulses are controlled in position and width under voltage control. Output of these modules are gated together in the binary "geometric region processor".
5) An array of binary functions called an "octal geometric region processor." A collection of eight digital functions of two signals: A and B, A or B, A EXOR B, are used to combine the rectangular pulses formed by the Voltage to Position Converter modules.
6) A Video Outliner called a "geometric unit generator" generates lines and points. The outliner has a horizontal edge extractor formed through delay of the video signal, and "EXOR-ed" with itself. The extracted left and right edges is selected to pick off the leading or trailing edges of the image. These horizontally derived edges trigger a 1-8 line "monostable" to form a rough approximation of vertical edge.
7) A Dual Video Processor - with gain and a "threshold control", to "core" out, and truncate video signals below a certain level. The processor can alternately be used as a level converter to translate audio signals to DV# 1 levels. This concession allows camera images to enter the direct video data path.
8) One Quad Mixer module - with 11 input patch connectors. Four front panel thumbwheel switches assign the patched signals from the pattern generators to one of the four color channels labeled A,B,C and D. Each of the four channels has a "gate" input to "turn-to-black" or turn off the signal with a video speed control voltage. Switch #0 is connected to a flat color field, switch #9 and #10 are hard-wired for the two external camera inputs of the Dual Processor. Each of the four channels has a low pass filter to smear the image, called a "texture generator" and can be set to either a horizontal or vertical time constant. Each of the four outputs drive a master level control which wires over to the Color Chord modules.
9) Four Color Chord modules - These modules superimpose the Quad Mixer output into triplets of red, green and blue levels which drive amplifiers with non-inverting and inverting inputs. Each module is controlled by its own set of six knobs, the superposition of the signals appearing as "color chords". Three knobs are assigned to the non-inverting Red, Green and Blue amplifiers, and three other knobs to the inverting or "negative" side of these differential output amplifiers. The amplifier outputs are DC restored then passed along for final output to the RGB to NTSC Encoder. A 3M NTSC color encoder and Telemation NTSC color sync generator develops the timing and final video output for DV #1. A simultaneous monochrome and color video output are available.
The Video Weaver is a digital pattern generator involving a string of counters and a Random Access Memory (RAM) to hold and later retrieve a stored pattern. It can be viewed as an electronic loom, having a vertical warp and a horizontal weft. The pattern is programmed into the memory then "woven" onto the screen by a set of phase shifting counters that slide and shift their count sequence in time to the video raster. A cursor is available to write in the pattern, while various phasing and counter direction parameters are used to offset the scanning order of the resulting video pattern. It differs from a strict frame buffer design in that the counters that read the memory are not locked into a static scanning order, but drift and wrap-around as the raster progresses.
VIDEO WEAVER II
The video weaver was "reincarnated" in 1992 specially for the Ars Electronica "Pioneers of Electronic Art" exhibition. The reincarnation impliments the original 1974 digital design within two ASIC chips that replace the 60 original 7400 series TTL logic chips. The functionality is the same. The new Video Weaver implimentation utilizes PLCA chips, programmable logic cell arrays, to include all counters and logic within a single CMOS gate array of 3000 gates. The user control and interaction is via manual switches and patching. (Engineering of the new LCA is by Stephen Beck, with the assistance of Kevin Fischer and Dave Barr. Additional assembly by Bob Vanegas.)
A few words from Woody Vasulka:
Reproduced with permission, from Pioneers of Electronic Art (1992), edited by David Dunn.