I’ve always been fascinated with computers from long before my time. 1980s CG animation in particular is something I’ve always found oddly soothing. But a lot puzzles me about how it worked. How were things like [Carla’s Island](https://www.youtube.com/watch?v=CUloIyGMM2A) made as early as 1981? If I were working as an animator at the time, what kind of program would I have used to create things like that? Weren’t computers at the time only capable of simple pixel images at most?
In: Technology
Carla’s Island was made on a Cray-1 supercomputer using a process called raytracing. This was a state of the art supercomputer in 1981 when the film was made.
Raytracing works by having the animator define everything in the scene mathematically, including the shape/color/texture of objects, any light sources, and the location of the “camera”. Then the compute goes pixel-by-pixel and goes “What would the camera see at this pixel” and calculates every bit (or “ray”) of light that would interact with that pixel to figure out what it should look like. Then it goes onto the next pixel and does the same thing.
The movie in question did some things to cheat and make life easier for the computer. For example the wave motion was periodic, so that a lot of the frames are actually identical except for changes in the colors. This meant that the super computer only needed to computer 144 complete frames. Then a much smaller computer could loop through that computed data an adjust the few things that changed (mostly the location/color of the sun/moon).
Those images were animated one frame at a time in a supercomputer at LLNL. Computers in 1980’s were capable of working in 3D, just not in real time (I was at SIGGraph in 1980). Essentially you figured out each voxel (3D pixel) and color coded it. Then you rendered the 3D space to 2D with a camera algorithm. You printed the 2D image on a piece of paper and photographed it, or if you had one, directly to film. Then you developed the film and projected it.
What’s most “breakthrough” about Carla was that the water has a surface normal. This means that the moonlight can reflect off it based on the relative height of the water in two voxels. That’s why the second half of the film is so dark, to show off the awesomeness of that algorithm.
Latest Answers