booksfilminterviewsmusictechnology
Tony Apodaca and Flip Phillips
"The time of 'glitz value' for CG is way over."
Emru Townsend: Both of you have watched CG animation grow from a curiosity to a gazillion-dollar industry. We've had T2, Black or White, and Jurassic Park remind the general public that computer animation means more than ultrachrome flying logos.

Some people think the next phase of evolution is CG actors replacing live ones. Others think it just means special effects will get all the more sophisticated, so the suspension of disbelief in, say, the new
Star Wars will be raised a notch.

Having watched things progress from the "inside", where do you think CG will go from here?


Tony Apodaca: CG has turned into a "gazillion-dollar industry" for exactly one reason: there are now two classes of visual effects for which executing the effect on a computer is more cost-effective compared to doing it any other way.

The first, most "mundane", and by far the largest class, is image composition. Computers are now able to scan images, digitally manipulate them in reasonably simple ways (warp, paint out imperfections, composite) and output them back to film more cost-effectively than the old hand-paining, rotoscoping and optical compositing. The fact that the digital manipulations are in some ways "cooler" than the optical manipulations is incidental. The speed of mid-1990's computers and quality of mid-1990's CCD cameras (the guts of modern scanners) have finally crossed the cost curve of old-fashioned manipulations. Note that for really simple manipulations (say, just a straight-up blue-screen composite) this is not yet true, but the effects houses are going 100% digital anyway because it will be soon.

The second class is extremely fanciful large-scale fully-articulated creatures, e.g. the water-weenies, T-1000s, and dinosaurs. These are only cost-effective on a computer because the other potential technologies (robots, makeup, stop-mo, etc.) are very expensive themselves. A decent- looking robotic T-Rex costs several hundreds of thousands of dollars, and can't walk. Stop-mo armatures are less expensive, but amazingly slow to animate (which equates to many months of salaries for animators). Computer animation systems are now faster than models for complex fully-articulated motion (again, not so for simple motion), and CG models are less expensive to build than robots.

So, when you ask, "What's next for CG?", you really have to consider what's still really expensive for a movie-maker. Honestly, I think that modelling everyday humans is a long, long way off. Despite the fact that a couple of my friends have been quoted as saying it'll happen in less than 5 years, and even assuming that modelling and rendering a realistic-looking human was already demonstrated to be possible (which it most categorically has not!), I still say look at the economics. Human actors are plentiful and cheap. Only really big-name stars make enough money to even worry about, and you can't very well replace them with a synthoid, because they are the definition of box-office draw.

As you know, at Pixar, we are working on a fully computer-generated full-length animated feature film. We are doing this because we believe that computer animation is now cost-effective relative to cel animation. Or at least, nearly so. There are several other companies who have come to this same conclusion, either independently or perhaps spurred by our belief. Again, this is only true for really high quality stuff, because it is really inexpensive to put together animation if you are willing to accept Saturday morning quality stuff.

My belief/prediction is that the true "next frontier" for CG is (don't laugh) pyro. True Lies has tens of millions of dollars of exploding buildings, bridges, vehicles, etc. These shots are really dangerous, tricky and one-shot-at-it affairs. It is not yet possible to get these effects by computer, but once it is, I'm sure we'll beat their costs in pretty short order. Think of the savings on insurance policies alone! Remember, you heard it here first!

(Of course, true sticklers will remind me that we've already had the first CG pyro: exploding LA in T2. Most of that was miniatures, however, so it's just the first small step.)

Okay, Flip, what say?

Flip Phillips: Interesting... I wasn't aware that that pyro was CG... 'mazing. I have to say, if there's one thing that leaving production has done, strangely enough, is that it has opened my eyes to some of the other stuff going on out there. Seems that when I was mainly doing production I was always caught up in whatever it was that we were doing, hell, I didn't have enough time to worry about other people's stuff... when a friend of mine showed me some of the footage in Jurassic that was CG my only reply was that of the proverbial deer in the headlights. (I have first hand experience with this phenomenon, by the way, but that's for me and my insurance company to deal with...)

I think Tony hit it right on the head with the price/performance distinction. One has to think rationally, and the time of 'glitz value' for CG is way over. In fact, back in the early '80s people were doing drawn animation so it would look like CG, since it was cheaper to do the drawn stuff and the 'glitz' of CG was 'in'. Go figure...

We see a lot of 'technology speak' that goes this way: 'photorealism' (to quote a friend and former Pixar colleague/photographer par excellence, Craig Good, "If you want photorealism, use a camera."), 'virtual reality' (virtual reality=theatre), now we're getting into this 'synthetic actor' thing. It is an interesting proposition but I can't really think of it as being really viable. As long as DeNiro or some lesser-paid actor is willing to sit in a chair for 12 hours having makeup applied we'll probably see humans, they're cheaper...

Of a little more interest is the strange practice of bringing people back from the dead. Kennedy in Forrest Gump, and the whole 'dead folks' Coke ad. I actually have a little bit of trouble with this. I guess I'm not too keen on the idea that 1000 years after my death I could be re-animated for a commercial like: "I'm not alive anymore, but if I were alive, I'd use BlatCo animation software!" complete with a talking, now dead, me. I wouldn't want my family to have to deal with that, even if they said, "OK, use his image, it's alright with us."

I need to remember to talk to my lawyer about getting that into my will.

I guess it's interesting now, but I doubt that we'll see much of it in the future. Faddish... now, the technology that went to making that stuff possible is really where the big win will come from. Computer graphics/animation is a little like that, we come up with a cool way to enhance satellite pictures so that we can drop bombs more accurately, and we apply it to cleaning up scratched old movie prints... My old mentor Chuck Csuri pointed out that, in the early days of the Computer Graphics Research Group, we had to take on government-defence-oriented grants so we could afford to make art. We'd always steal stuff from those projects, image processing for medical images became a way to make interesting painterly effects with cold sterile images...

As for where it's going, it'll probably go in split directions, the glitzy stuff will be around at a high price, just 'because it's CG' and the economical stuff will pay the bills. I think the Pixar film will get a lot of press for being all CG. I'll guess that the attention to that fact will be a bit misguided, since no doubt the story and quality of the animation will be so high. But people in general are really tied up in the 'toys' with an almost geek-like fascination. Hopefully people will see past the shine of the method and look at the product.

Are you guys working on anything right now?

FP: A little... like I said earlier, I'm currently doing my Ph.D. in cognitive psychology, doing vision research. Surprisingly, computer graphics and visual psychophysics have quite a bit in common. This has obviously come in handy. I'm also doing some aesthetics research as well. I have some fun systems that I have been trying to use to measure and model aesthetic preference. Fun stuff. It would be nice if it could be integrated into design software, perhaps to provide some type of assistance in exploring the rather substantial solution space of design problems.

I also have been doing a little work on the side for Ion, a second generation spin-off from Pixar. We did the new David Bowie CD-ROM, along with Brian Eno's and The Residents'. We're pushing hard to create the next generation of music album. CD-ROM is an interesting delivery medium. Right now it's rather platform-specific but new tools and emerging standards might change that in the future. I don't think it will go the way of the proprietary video game market.

TA: As you no doubt know, Pixar is currently producing the world's first fully computer-animated and computer-rendered full length motion picture. The title is Toy Story, and it stars the voices of Tom Hanks, Tim Allen and several other famous and familiar actors. It will be distributed by Walt Disney Pictures, and will be out for in time for Christmas. We cannot talk much about the details of how it is being produced, and we really will not be able to until next year (after the film is actually released). However, I will guarantee that the computer graphics and computer animation community will absolutely flip over this film. Not only are the images themselves totally beautiful, but the story is really exciting and funny. This is not gonna be a repeat of Tron. This is, of course, what Pixar specializes in: writing very compelling stories with real believable characters, and bringing them to life using the medium of computer graphics.

page 1 | page 2 | page 3

Originally printed in fps #7 (Autumn 1995)
Feed the hungry every time you search the Internet! Click here!