SIGGRAPH 2003: July 30
Emru Townsend · July 30 | On Sunday I made an offhand comment that I would be chairing some sessions today. Let me explain a little.
Earlier in the year, people were invited to submit proposals to the Sketches and Applications committee ("Sketches" for short). Sketches are designed to be a showcase for new ideas and techniques that have recently been established, or are being introduced. (Sketches aren't as formal or as rigourous as academic papers, so they're becoming more popular for both presenters and attendees.) The Sketches committee is divided into several sections, each with its own subchair. This year I was the Animation subchair. (Last year, it was Dave "Grue" DeBry—who also happens to host this website.) What that means is that it was my job to pick the three other people who would make up the Animation jury and help guide the process by which we would sift through the year's proposals and decide what would make the cut.
At the end of the jurying, all the accepted Sketches from each section are looked at together, and divided up into four or five-Sketch sessions according to theme, and each juror gets to chair at least one session at SIGGRAPH. This year I chaired three sessions—and, as it happens, all three sessions were back-to-back on Wednesday. (That's what happens when you duck out of the scheduling meeting early.)
The first session was Building Character, which looked at the challenges of creating believable characters that would interact with live-action in some way. The presentations involved work on Lord of the Rings: The Two Towers (Weta Digital), the "Jordan vs. Jordan" Gatorade commercial (Digital Domain), The Hulk (Industrial Light + Magic) and an NFL commercial (RhinoFX). My personal favourite was the Vico Sharabani's presentation for RhinoFX, probably because it was the most elegant. The ad in question used an X-ray effect on footage from hockey games, with pistons and other engine parts shown to be part of the players' skeletons. Being a smaller studio (compared to the likes of ILM and Digital Domain) and under TV-commercial time and money constraints, Sharabani didn't have the luxury of fully animating a skeleton model in 3D space. So he did what artists and animators have been doing since time immemorial: he cheated.
Rather than animate the skeleton model in 3D, Sharabani simply made key joints on a skeleton (say, the shoulder, elbow, and wrist for an arm) match up with the corresponding spots on the live hockey player's body. This distorted the skeleton in 3D space, but looked at head-on it was nearly perfect. Then it was simply a matter of adding the X-ray effect, compositing, and so on. You could almost hear the crowd's eyes pop open upon hearing of such a simple and effective solution that produced such excellent results.
Two presentations in this session are of particular interest to fps readers. Mainframe Entertainment's Gordon Farrell explained the evolution of the cel-shading techniques they used for, among other things, this season's Spider-Man and the ill-fated Gatecrasher (based on the Mark Waid/Jimmy Palmiotti/Amanda Connor comic). I mention these two examples because they were the most stunning of everything Farrell showed. In particular, Gatecrasher perfectly nailed the look and tone of the comic book, and I'd have loved to see this kind of work on a regular series. But MTV and Gatecrasher ultimately parted ways, and the music network instead picked up Spider-Man—which, quite frankly, looks awesome. I've always liked the concept of cel shading better than the execution, but the extra effort put in for details like consistent linework seem to have really paid off.
The last presentation in the session had a mouthful of a title: "A New Form of Traditional Art—Visual Simulation of Chinese Shadow Plays." I-Fan Shen showed how he and his colleagues were working to preserve the ancient but dying art of Chinese shadow plays digitally—by replicating the puppets and projecting light through them using off-the shelf 3D animation software. But rather than render animated images of the puppets, the "camera" is instead pointed at the resulting shadows. The result is instantly recognizable to people who have seen Lotte Reiniger's silhouette films, though the images are quite a bit softer. Shen deservedly received a considerable amount of applause after his presentation, which goes to show that the SIGGRAPH crowd, which can sometime seem too high-tech for its own good, can appreciate aesthetics as well as any other.
My final session for the day was The Matrix Revealed, which looked at techniques used throughout The Matrix Reloaded for its virtual stunt doubles and sets, but focused mainly on their application in the Burly Brawl sequence, where Neo (Keanu Reeves) battles hundreds of Agent Smiths (hundreds of Hugo Weavings). While moderately technical—if you don't know about lighting for 3D you might have gotten a bit lost—it was generally fascinating to see how ESC Entertainment tackled problems like rendering and animating convincing hair, skin, and cloth. There was also a great moment where they showed part of the Burly Brawl with Entertainment Tonight host Bob Goen's smiling head rendered in place of Keanu Reeves's, which was great fun (I hope it makes it on the DVD.)
ESC's George Borshukov (to whom I apologize again for mispronouncing his name, though he got immediate payback by walloping me on the shoulder) answered two questions which I thought were important for animation and effects fans.
First: Is it really so important to put all this effort into subtle details that zoom by too quickly for the eye to register? (They had to re-render some movie stills without motion blur in order to point out some of the hair and face details, for example.) Borshukov's response—I'm paraphrasing here—was that the attention to detail, even if not consciously registered, is what leads to a scene's believability. It was sort of the philosophical opposite of Vico Sharabani's "cheating," but I can see the truth in both approaches.
Second: How did the actors respond to being replaced by digital duplicates? Referring to his earlier examples, where Reeves, Weaving, Laurence Fishburne and Carrie-Anne Moss had to act out various emotions to register facial animation, Borshukov replied that they didn't see it as replacement at all—in fact, they loved how much they were included. He'd also mentioned earlier that the better the performance the actors were able to give, the better the results the ESC team could produce. It brought to mind the cliché that so many writers use when discussing animated characters—that they don't have the hassles of "real" actors. But the people who bring animated characters to life, from voice actors to animators, are actors, and just as in live-action, the talented ones will produce the most compelling characters. It was good to see someone working on a movie with extensive use of virtual characters acknowledge that.