A Hybrid Analog & Digital Animation System

Dave Sieg

Image West had been in business several years when I first came there in 1980 as a maintenance engineer. The company had been bought from Computer Image's bank at a bargain, and had been operating successfully for several years. They were in the midst of splitting from a Canadian parent company, called Omnibus. (John Pennie, the other owner of Omnibus went on after the split to form Omnibus Computer Graphics in Toronto.) Cliff Brown, (shown at left in an image captured from videotape) who was then president of Image West, asked me to recommend some options for future development directions. In 1981, there were few options. Digital image-making state of the art was a PDP-11 and a $50,000 framebuffer, and a bunch of assembly or FORTRAN programmers hacking away from scratch. Triple-I, NYIT, and MAGI were about the only people going that route. Image West had always had the advantage of "Real Time", meaning that despite the limitations of the analog rescan technology, it could run right before your eyes, and be adjusted on the fly. Its big downfall was complete lack of repeatability, due to all those knobs and patch wires. After reviewing all the options, Cliff and I decided a good approach would be to build a system based on the analog rescan technology, but using digital computers to track and store the setups needed to repeat a job. I did not realize at the time how large a project this would be.
Since one of the weakest links in rescan technology is the CRT, I built a test jig to evaluate and test CRT performance. The Scanimate's CRT systems had shortcomings due to lack of resolution, and their magnetic deflection limited the bandwidth to a hundred Khz or so. We had a prototype 3D rotation matrix that would work in real-time, but it required the deflection system to run at up to video speeds. So, we had Special Purpose Technology Co. design us a sub-screen CRT (not the one shown here). The reason CRT's tend to have such low resolution is due to the light from the spot bouncing back through the faceplate, which is usually about 1/4" thick to hold the vacuum. If you analyze the spot through a microscope, you can count up to seven rings around it from secondary internal reflections off the faceplate. Some modern film recorders fix this by using neutral-density faceplates, or in some cases coupling the faceplate to another thick faceplate with a laser liquid bonding material similar to mineral oil. The SPTC CRT had a specially designed fine grain phosphor deposited on a thin piece of glass (the subscreen) that actually sat a half inch behind the actual faceplate. Thus, the resolution was excellent, because the rings were nonexistent.

Here is the electronics racks that made up the final version VersEFX. The rack at the right held the "master" system, which talked to each "channel" via a IEEE 488 buss. We should have used Ethernet, but in 1981, it was less available. Jim Ryan did an amazing job of writing the assembly code that ran in each system, dealt with realtime updates, playback of animation, and all those lovely I/O registers and their quirks. You can see an ICE (In-Circuit Emulator) probe sucking onto the master processor if you look closely.

Here is a closeup of the CRT section of one bay in the final version. If you look carefully, you can make out the subscreen CRT. Rather than the awkward protruding camera like Scanimate had, I chose to fold up the whole optical path using front-surface mirrors. The whole thing fit into a rack-mounted slide system, and plans were to introduce filtration mid-path at some later point. The CRT was scanned by a COHU camera, which turned out to be a very poor performer. Later plans were to take an old studio camera and split the individual red, green, and blue electronics out to allow registered RGB signals to be animated. Alas, advances in digital framebuffer technology just did a better job of storing an image than the face of a CRT. Also, as it turned out, our high-resolution CRT was TOO high-res! The Scanimate had used that big fat point to fill in a lot of space between scan lines.

This was the VersEFX control console, manned here by Jim Ryan, the project software engineer. The bay came out very much like the original artists rendering above. Unfortunately, despite having shifted all costs over to the SFP, Image West ran into financial difficulties, and the only system built was shipped to France. Peter Koczera claims he went over and that the system was working and they did some interesting things with it, but it was never completed, partly because the design goals were moving targets throughout its development. I learned a lot about managing projects of this size, most specifically how fast technology can change, often before a big design can be completed. It also became apparent how quickly a small company can be swamped trying to do R&D on this kind of level.

All images Copyright © Dave Sieg Back to Dave's History Page
Please email me for more info: