Beowulf... Why?

Robert Zemeckis, a very talented filmmaker, is back this season exploring the highly unpromising vein he started mining with The Polar Express. Beowulf is a loose retelling of the Old English epic poem. Like Final Fantasy, these Zemeckis films are photoreal depictions of human characters. The innovation that Zemeckis has clamped onto is procedural; rather than animating from scratch, he records actors’ performances, this go-around with a new motion capture technique that photographs them from multiple angles. I saw the rig he uses for this at the SIGGRAPH convention this summer; it looks like a geodesic dome fitted with dozens of lights and cameras.

Despite my misgivings about motion capture, I came to the film wanting it to succeed. I saw it on the first weekend, and in IMAX 3D. Unfortunately, I have to report that the film was an almost complete failure. Only two scenes really worked: Angelina Jolie’s seduction of the hero, and the final action set piece. For the most part it was a flat, frigid bore.

The question I was left with was… why? If this film had been made in live action with CGI assistance, it still would not have been a good film, but it would have been a better film. I know what Anthony Hopkins and John Malkovich can give a story. What motion capture got out of them was barely 10% of what they can do even when they’re having a bad day.

What went wrong? First, the motion capture doesn’t look that bad. When it’s bad, motion capture looks… dirty. There’s more detail there than needs to be there. Zemeckis has obviously had animators edit the motion extensively. The trouble is not the usual one of too much extraneous detail, but not enough. There’s no intimacy to the performances.

When humans look photo-real we expect photo-real expression. The human face is a highly complex interweaving of dozens of muscles. Moreover, we are extraordinarily adept at reading it; our species has evolved into master interpreters of our own faces. Small subtleties carry enormous meaning. The fault might not be with the motion capture alone, but the character rigs too. 3D technology has advanced enormously, but it still cannot quite simulate all the details of facial movement, at least not when the benchmark is human complexity.

Perhaps it is also the circumstances in which the motion was captured. An actor is far likelier to create a compelling performance on a set, with the actors he’s playing against right there in front of him. A geodesic dome is not quite the same spark to the imagination.

Finally, there is the issue of what 3D folk call the “Uncanny Valley”. This term was coined by MIT researchers on robotics. They were trying to come up with a robot design that would provoke emotional attachment. They found that the more like humans the robots got, the more warmly people responded to them. This perhaps was not surprising. What was more surprising was that there came a point when this effect started diminishing and reversing. People started getting a little freaked out by the human likeness. When the design started becoming completely convincing – like say, Rutger Hauer in Bladerunner – identification and attachment returned. This effect was dubbed the uncanny valley. People in 3D argue about whether it really exists. As one who has fallen in myself (in a mixed media project,) I can testify that it does exist. Does Beowulf fall into the uncanny valley? Well, it’s clawing up the far side, but it’s there.

I go back to the most pertinent question: why? Why bother recreating reality in mathematically defined polygons, when real reality is right here with us? I know the answer in my bones because I’ve made films both in 3D and live action. A live action shoot is terribly frustrating for control freaks like me… and presumably, Zemeckis. A plane flying overhead ruins your best take. The damn cameraman muffles the follow on another good take. The actor scratches his nose for no apparent reason, ruining yet another one. And if you say you’re satisfied with the shot, at 1AM in the morning, when you’ve been up and working your ass off since 6AM, then you will have to live with that for the rest of your life. You rarely get retake days. The 3D world, on the other hand, is infinitely malleable. You can tweak the camera angle or the hue of the hero’s plaid jacket till the cows come home. Yes, there are time constraints because budgets are never infinite but things can be… adjusted. Always adjusted.

The truth, however, is that this kind of stuff doesn’t really matter all that much. Filmmakers can obsess about things intricate like camera movements or subtle pictorial elements, but that’s just the icing on the cake. What matters is story and performances. To sacrifice performance for the flexibility to execute the filmmaking flourishes is terribly, terribly wrong-headed. And I just can’t imagine what other upside there is in a technique such as this.

I believe 3D animation does not need to be confined to family audiences. It can appeal to older, narrower audiences. But I also think animation needs to be allowed to be animation. Let it do what it does best: provide an imaginative restyling of life movement and life imagery.