I was first introduced to stereograms when I was a young child, my family would visit my Grandmother and she would always bring out her Victorian era stereogram viewer and old stereo photo cards showing exotic sights from around the world. Perhaps this was a bit ironic since my Grandmother was totally blind. She became blind as an adult, so perhaps she was sharing the same images she viewed as a child.
I learned to draw stereograms by hand in a perspective drawing class in my first year of art school, an arcane procedure that confounded most drawing students, but I enjoyed it immensely. Each year, a collector of stereograms came to the art building and set up a stall to sell individual stereo photo cards, I used to spend hours looking through his collection, but since I was a starving artist, I never had enough money to buy any of them. When I first got access to 3D computer graphics hardware in art school, my first project was to produce stereograms. Unfortunately, the hardware was primitive and low resolution, and the depth effects were difficult to perceive. I remember how difficult it was to get my professor to come over to the art school and see my first stereograms, since the only graphics terminals were in the Computer Science building, and artists of that time wanted nothing to do with computers. I finally managed to get him to come over for a demo, I produced stereograms of bright arcs swooping through space, an homage to a famous sculpture by Alexsandr Rodchenko. The professor devastated me with the remark "oh, that's just technical." That was when I decided to drop out of art school. It would be 15 years before that professor jumped on the bandwagon, and tried to make his reputation as a guru of 3D Virtual Reality. And of course, by then he had completely forgotten that I was the first artist to ever show him a stereoscopic VR image.
After dropping out of art school, it would be amost 10 years before I could afford my own computer equipment capable of rendering stereograms. I did some primitive stereoscopy experiments with Paracomp Swivel 3D, but the first application capable of doing the job properly was Specular Infini-D. One of the demo images in the Infini-D application was an animation called "virtual gear." I viewed it and was immediately irritated, it was just a disk spinning on its axis, not a gear at all.
Now if you're going to call your demo "virtual gear," you have a historic precedent to live up to. One of the most famous interactive computer graphics experiments of all time was Dr. Ivan Sutherland's Virtual Gears. Sutherland set up an interactive graphics display showing two gears, you could grab one gear with a light pen and rotate it, and the second gear would move, meshing teeth with the first, moving as perfectly as real gears would.
Gears are a graphics cliche that annoys me continually. Gears are overused as an iconic image, and are almost always badly designed. I can't count how many times I've seen animated gears that would not work in reality. The gears don't mesh, or are combined in mechanisms that would jam. I've even seen gears that counterrotate against each other, if they were real gears, all the gear teeth would be stripped away.
My Grandfather was an amazing tinkerer, and was always constructing clever little gadgets with gears and pulleys. My favorite gift from my Grandfather is a college textbook from the late 19th century entitled "Principles of Mechanism." The book shows how to design gear and pulley mechanisms, and shows how to distribute power throughout an entire factory from a single-shaft waterwheel. I always enjoyed the strange diagrams and mechanisms, and was astonished how much advanced calculus went into the design of even simple gear teeth. So when I saw the Infini-D image, I immediately thought back to "Principles of Mechanism" and decided I could do better. I would design a properly meshing set of gears that was accurately based on real physics. To make it a challenge, I would make the gears different diameters with a gear ratio other than 1:1. And to add realism, I would produce it as a ray-traced 3D animation. Click on the image below to see the finished animation pop up in a new window.
In order to design the gears properly, I had to sit down and work through quite a bit of the physics of gear design. I discovered some interesting things about gear ratios, if I designed two different diameter gears with the right ratio of teeth, both gears would return to a symmetric position after only a half rotation of the largest gear. This would allow me to produce a fully rotating animation loop without having to animate the full rotation, I could just loop the same frames twice, saving half the rendering time. Look closely at the animation, the slots in the two gears match up after the left gear rotates only 180 degrees. There are 60 frames in this animation, but it takes 120 frames to complete one full cycle. I saved 50% of the rendering time.
And it was good thing I discovered this trick, the animation took 48 hours to render on my Macintosh IIcx. I borrowed a Radius 68040 coprocessor board and the animation rendered in about 24 hours. When I bought a new PowerMac 8100/110, the render took about 20 minutes. I haven't benchmarked this file since then, but I suspect it would render in about 30 seconds on my current dual 1Ghz G4 computer (if I could get Infini-D to run at all).
My stereoscopic rigging in Infini-D took months of fine tuning to get it to work perfectly. About that time, Specular Inc. started releasing inexpensive accessory packs for Infini-D, selling for about $100. I had previously contributed some work to Specular, producing a procedural texture generator for cloudy skies, the "Sky Library" which you see in the background of this stereogram. Sky Library was Specular's most popular software download, and they knew of me and my work, so I started negotiating with them to sell my camera rigging as a software accessory pack, with an inexpensive plastic stereogram viewer included. I sent them several animated stereographic demos including Virtual Gears, high-resolution Iris inkjet printouts of various stereograms (which were very expensive to produce in 1992), and a freestanding stereo viewer. They had never seen stereograms and did not understand how the viewer worked, they tried to hang the metal legs over their ears, so I sent them a videotape showing how it stood up on the metal legs. I worked for months, trying to get their interest in releasing my work as a product, but they told me there was no market for my rigging, and they didn't want to release it. I persisted, but eventually they stopped returning my calls.
A few weeks later, I received a mailing of the Specular Infini-D users' newsletter, and was astonished at what I found. The newsletter's feature article was about setting up stereographic camera rigging in Infini-D. The article used all the methods I had taught them, but the setup was all wrong. They had taken all my careful design, published the core concepts, but missed all the mathematical subtleties that made the images work as a stereogram. And to add insult to injury, I did not get any credit at all, they claimed they invented it all by themselves.