Visual Effects on Terminator 2

Alex Seiden, a senior technical director and animator at Industrial Light & Magic gave a talk about Terminator 2 and its visual effects at the International Animation Festival Cardiff. Report by David Jefferson.

“Behind all magic there is a trick, and it is certainly true in the case of visual effects. It is not really magic it is just a bunch of stuff we can do that makes things look good and makes things look right Today I am going to show you how some of it is done,” said Alex Seiden in his opening remarks to a packed audience in the main theatre at St David’s Hall, Cardiff. As senior technical director and animator at Industrial Light & Magic he is supremely qualified to explain the technical background of the hit special effects film Terminator 2.

A grid on Robert Patrick helps as a reference to creating the T1000.

Terminator 2 was probably the most ambitious computer graphics film since Tron. Tron had great effects in it but it wasn’t very successful at the box office. Tron was never supposed to be real, you were inside a computer. Similarly, with the Genesis sequence in Startreck 2 Kirk was watching a videotape.

In Terminator 2 there are real live action characters interacting with computer graphics. You are not supposed to think of them as effects or some magical, mystical thing, you are supposed to think of them as something tangible that you can touch.

An earlier film, The Abyss, was a great landmark for ILM and we won a Oscar for it. However, the director knew that if the effect didn’t work he could cut it, he would have found another way of doing it.”

In Terminator 2 they were animating the title character. and the success of the movie depended on their sequences. The effects ILM created for the TI 000 character could only have been created with computer graphics. The nuclear sequence and the future world sequence was created with traditional special effects technology. They used things like miniatures, optical compositing, practical gags. puppets on the stage and so on. ILM had to ensure their shape changing computer graphics matched exactly to live action footage seen twenty frames earlier.

The computer graphics took approximately one year from initial boarding stage to the final delivery date. It was ten months of actual production for thirty-five people so about twenty-five man years of work went into a computer graphics sequence that makes up less than five-minutes of running time in the film.

Seiden identified the five basic steps in producing the animated computer graphics, although some of them overlap a bit: modelling, animating, lighting, rendering and compositing. “Modelling is telling the computer what the geometry or shape is. Animation is telling it what the motion is. Lighting is telling it where the lights are, for simulating different light sources, different reflections. how shiny something is, whether or not it has shadows and so on. Rendering is what the computer does, it is more like waiting because it can take hours or days depending on the computer power and the complexity of what you are doing. Compositing is mixing computer graphics with live action and that is done with ever single shot we have at ILM. There is no point in duplicating Arnold Schwarzenegger on the computer when we can add the computer graphics to the live action.”

People working with computer graphics are found sitting in front of a work station for hours on end, with a mouse, dragging points around, fitting things in, changing key frames and making it happen.

The modelling step involves tracing the shapes, the outlines and the contours of the surface they are building in a number of separate views.

Animating on a computer involves setting the key frames. They use wire frame work which is the computer graphics equivalent of a pencil test. The kind of decisions the animator is making on a typical scene, will be deciding where to place a cross section, or how a spline should move.

At the rendering stage there are still some details to be animated. However, some things such as the ripples on the surface of the water are better left to the computer. To do that kind of animation by hand would be really tedious because there are so many points on the surface and great detail is required. “We can do things procedurally by describing to the computer how water moves, displace the surface of the water with a sine wave that propagates through. Although it may sound like the computer is doing it for you it is not really the case because you are telling it how it does it. You tell it how fast it does it, whether it is high frequency like liquid or whether it is more like jelly, or mud, or chewing gum. The animator has to know how things move and how to describe that motion to the computer.”

page 1 | page 2 | page 3