John Dykstra interview pt. 4

INDEX: On the Dykstraflex | on ILM | on CGI | on Hancock | on Spiderman | on Stuart Little | on Batman And Robin | on videogame development | on future projects | on favourite SFX films | favourite own SFX shots

Continued from part 3

We've been talking a lot lately about how impossible camera moves can make good CGI look fake, and yet at the same time artists are putting back real-world limitations such as camera shake, to 'ground' the shot in available technology. You yourself invented a virtual 'spider-man camera-operator' for Spiderman's CGI, so do you think that kind of visual limitation is necessary?
We were trying to make cameras do impossible things before the advent of the computer. We managed to make a lot of that happen by varying film speed, using different photographic techniques, miniatures, maxitures, time-lapse photography, weird lenses...you name it. So we were cheating even then, there's no question about it. I think the concept of camera moves is an important one. I don't like to make camera moves that are impossible in and of themselves. I like the idea of doing something in a camera-move that's impossible, but I like the basis of the camera moves in general to be from reality. An advantage with me of coming from the photographic world is that I know what you can do with a camera.

So what I set about to do was not to design an impossible camera move. Spiderman's a good example: we're gonna create a cameraman for Spiderman, and the paradigm for this is when they used to do sky boarding, where a guy'd leap out of a plane with a snow-board attached to his feet, do a bunch of aerobatic manoeuvres and land, with a parachute. The competitors were always teams; there was always the guy who did the tricks, and the cameraman, because the judges obviously didn't jump out a plane with him and couldn't see that well from the ground.

So this cameraman went along, and his performance was also interpreted, because the judges viewed his film to see how well the performer did. So it became an art in itself to not only stay with the guy while he's doing all of these manoeuvres, but also select the best angle, keep the sun at the right position, make the thing look good.

So I used the sky-diving cameraman as my paradigm for the virtual camera-moves that we did. So I said 'Okay, Spiderman's got a buddy, and the buddy wears a camera, and he has to be able to do Spiderman-type moves while focusing the camera on Spiderman. So for most of the stuff, you can imagine the guy who's with the camera swinging along past Spiderman…we accelerate past him or we go below him or we go above him, but we don't jump there; we don't go there too quickly.

We do it as you would do it if the camera were actually attached to something physical. Now, add to that the part - which I like - which is the 'gee whiz' factor, where we go through the rails of a fire-escape, or we go through one window and come out another - things that you couldn't do if you were actually having a guy be a physical cameraman. But those are accessories and enhancements, rather than the basis for the conceit for how the camera moves. So you get the little 'gee whiz' stuff, but for the most part the idea is based on something that would be a 'real' thing.

Having said that, you've got to remember that movies are always cheats: when Spiderman swings through the city, if you were actually to figure out his physical speed, he's travelling two hundred and fifty miles an hour, which you don't do if you swing. If you swing, you hit fifty-five, sixty miles an hour under the arc. Maybe. But when you do that it makes for a very leisurely swing through the city [laughs], and it's kinda not as exciting as it should be.So we make Spiderman go a lot faster and, of course, we make the camera go a lot faster. But the camera and Spiderman and the movement through the city is limited to their own world - their enhanced world.

Another example: we did a lot of motion-capture to try and figure out if we could get real physical people to do some of the manoeuvres. Well, first of all, almost nobody is strong enough to do Spiderman-type manoeuvres; the whole idea of being able to pull yourself up to where your fist is below your ass and your legs are pointed straight out in front of you…a rope climber can do that occasionally, but the stress of trying to do that under a g-load is almost impossible. So we did, as a perfect example, motion-capture of a guy jumping off from the top of a platform and then landing on the ground, and the idea was to have the CGI Spiderman to use that motion-capture to show Spiderman leaping off the top of the building and then extend the part in the middle and then use the motion capture of the landing for the impact on the ground.

The problem is that the stunt-man's only capable of jumping his own height before he starts getting hurt, and the body-position and the deflection that he has when he hits the ground...the stunt-man would never tell you that he's human, but he is [laughs]. They just looked like they were at the limits of their strength landing at that slower speed, so when we sped the whole thing up, it looked improbable. It looked human. Spiderman is superhuman. So we had to create an animation that showed a different application of musculature, a different centre-of gravity, a different rate with which his body changed position when he impacted and the way he absorbed the impact.

In this case the reality, which is what the motion-capture gave us, gave itself away. It didn't look right. You'd look at it and go 'Oh, he just died!' [laughs]. That's what you have to constantly do - balance. Movies are storytelling, and when you're telling a story, you amplify some things and you leave some things out.

Part 5: Batman and Robin...and happier matters

Part 1 - Part 2 - Part 3 - Part 4 - Part 5

Complete list of the DoG Clone Wars reviews