The animation directors on Peter Jackson’s King Kong were Christian Rivers and Eric Leighton.
I mention this because from all the media coverage, you might assume that Andy Serkis was the single-handed creator of the character of Kong, just as many sources suggest that he was the single-handed creator of Gollum in Lord of the Rings. It has been suggested, for example, that Serkis’ work on both characters was worthy of Oscars for Best Supporting Actor (and, indeed, Serkis did win several acting awards for Gollum, as listed on his website). Yet, as should be obvious, Serkis is not the sole creator of either performance: both Gollum and Kong represent a blend of the performances by Serkis and the various animators at Weta Digital. Even an article as informative as this one at ComingSoon.Net – which does discuss the split between Serkis’ work and the animators in some detail – is based only on Serkis’ account and runs under the headline “Andy Serkis IS King Kong.” And of course the credits of the movie include a credit reading simply “Andy Serkis as Kong.”
I’m not trying to have a go at Serkis here. I think his collaboration with the animators at Weta on these characters is extremely important and groundbreaking work. Gollum, in particular, pushed the level of acting and psychological complexity we expect from digital characters, and exposed the mediocrity of the work on Jar Jar Binks and Yoda in the new Star Wars films. And I think an Oscar could easily have been justified for that work: I just think it should have been a special achievement Oscar, awarded to the entire crew who worked on Gollum, rather than just Serkis.
It’s a shame that this oversimplification has occurred, as the Serkis / Weta collaborations fundamentally reshape the way we think about the relationship between live and animated performances. Live reference footage has been used since the days of the earliest hand-drawn animated features: the animators on Snow White (1937) used reference footage on Snow White and the Prince, for example. Animation can be directly derived from the footage by tracing it, using a device known as a rotoscope, but the conventional wisdom has been that over-reliance on rotoscoping leads to poor animation. Better animated “acting” is achieved where the animators used the reference footage as a jumping-off point only, and adjusted the characters so that they gave a less literal performance: films such as Ralph Bakshi’s animated Lord of the Rings (1978) were full of dull rotoscoped animation, and the results were unsatisfying.
In computer animation, the equivalent is motion capture, where the performance of a live actor is captured digitally, and this has been the subject of a similar stigma. It isn’t, as far as I can tell, heavily used in the creation of Pixar’s films, for example, as they use relatively cartoony designs and consequently need a different type of performance. (I suspect, however, that it was used quite a bit for human characters in the Shrek films). Motion capture has been associated much more with animation that interacts with live action footage, as in the Star Wars prequels, and here the naturalness of its motion is of more obvious value. There has been a divide, then, between what you might call the special effects tradition and the cartoon tradition in computer animation. The literal realism of motion capture has been highly suited to the former, while the latter has tended to rely more on animator-driven performance.
The most notable predecessor to Gollum, Jar Jar Binks from The Phantom Menace, tried to marry the two approaches, but without much success. Much of the animation of Jar Jar’s body was derived from the performance of Ahmed Best, but the character’s facial features were so removed from a human’s that his performance could only be used as reference footage for the facial animation. What’s more, Best didn’t try to give a natural performance: his walking action for Jar Jar, for example, is that of a human trying to walk like a cartoon character. Jar Jar offered no opportunity for psychological complexity, or a performance of any but the most basic sort. This is why Gollum was so much more notable an achievement. He was a psychologically complex character: indeed, he was effectively two characters, as his mind had been fractured into the the evil Gollum and the (comparatively) good Smeagol. It was the first time someone had entrusted a digital character with such a key role in the movie, and the robust performance was an impressive blend of motion capture and animator-driven techniques. Between them, Serkis and the Weta crew had managed to combine the expressiveness of an animated performance with the realism of motion capture. Old assumptions about the merits of motion capture were thrown out the window, and the character showed that it really was possible to combine the best of both worlds.
The title character in King Kong was in some ways even more challenging. There wasn’t the crutch of dialogue to express the character’s thoughts and feelings, for starters, and as the depiction is of a recognisable (albeit exaggerated) real-world animal, the audience has more of a real-life reference to judge the character against. Yet the final result is if anything even more impressive than Gollum: while not as revolutionary, Kong is even more realistic. As I said in my review of the film, he is always gorilla-like, while at the same time the range of and complexity of emotion he displays is quite startling. The performance has revived the interest in these groundbreaking digital characters, but once again, there has been a tendency for critics and journalists to fudge the story by assuming Serkis = Kong.
Yet Kong is arguably less Serkis’ performance than Gollum was. There isn’t as extensive a vocal performance, for example. And while Serkis did provide some on-set performance reference for Naomi Watts, given the difference in size between actor and character, this could not have been as tangible an interaction as occurred with Serkis, Elijah Wood and Sean Astin on the set of The Lord of the Rings. Finally, there is a greater difference in physiology between the giant gorilla in King Kong and the essentially humanoid Gollum: there would have been more need for animators to adjust Serkis’ performance, simply because Kong’s face and body are constructed so differently. VFXWorld (in a very interesting article I can’t link to because it requires registration to view) states that only 25% of Kong’s facial animation was derived from Serkis. While that may well have been the key 25% (such as the scene where Ann wins Kong over), even this animation must have taken a fair bit of “finishing.” If you look at the facial animation on Kong, particularly when he is angry, you’ll see his expressions are well beyond what a human face can achieve.
So the character of Kong is a collaborative exercise, rather than a star turn by one actor. Yet once the importance of that collaborative effort is understood, pointing this out doesn’t need to be a slight on Serkis. After all, pretending that a digital character is just like a costume that an actor wears – and I’ve seen this argued, in so many words – is not only insulting to the animators who co-create that character, it also distorts the understanding of what’s important about Serkis’ work. It suggests that all he’s done, conceptually, is a variation on the performance by Rick Baker inside a monkey suit in the 1976 version. Ultimately, being part of a team effort that revolutionises the way we think about animated performance is more significant than being the sole creator of just another convincing monkey.