Throughout the history of cinema, few monsters have remained as iconic as King Kong. Originally debuting in a 1933 film, the massive primate has been re-created multiple times using a variety of visual effects techniques: the painstaking stop-motion animation of the original film, the man-in-a-suit goofiness of King Kong vs. Godzilla, Andy Serkis’ motion-capture performance in Peter Jackson’s 2005 reboot. But when it came time for Kong: Skull Island, director Jordan Vogt-Roberts turned to Industrial Light & Magic to create a new version of the creature that could match his film’s shot-on-location aesthetics.
According to visual-effects supervisor Jeff White (Warcraft, The Avengers), the key to bringing the new Kong to life was a combination of old-school and new-school techniques: using keyframe animation to give the gargantuan creature the scale it needed, while drawing inspiration from motion-captured facial performances to add personality and humanity to Kong the character. I chatted with White to discuss the inspiration behind the new creature, the challenges of creating a 100-foot-tall ball of hair, and the key to making something like Kong more than just another special effect.
This interview has been edited and condensed for clarity.
Not a gorilla
There have been many permutations of King Kong over the last 80 years. When you talked to Jordan, what was his creative mandate for the creature?
Number one: not a gorilla. Number two: not something we've necessarily seen in previous Kong iterations, but certainly looking for the most inspiration from the original ‘33 Kong. That was really interesting, because as we started talking about the project, we did what we naturally do, which is pull lots of gorilla-movement reference [footage]. But after we had that first meeting with Jordan, we definitely were changing tracks a little bit, where Kong was not going to be a quadruped, and he was going to stand upright. Those decisions really influenced so many things about this Kong. One thing we were very worried about with him walking upright was, “Is he going to look like a guy in a suit?” We wanted to make sure that he still felt like this creature.
“For Jordan, it was really about getting back to this sense of a movie monster.”
For Jordan, it was really about getting back to this sense of a movie monster, rather than just a giant gorilla. We had some really fun discussions and things to solve as far as what his anatomy is, his underlying sculpt. He really is this interesting hybrid of man and gorilla. Gorillas can't actually stand up very well. It's very uncomfortable for them, because their ribs end up resting right on their pelvic bone. So for this guy, we re-sculpted him, kept the shorter legs, very long arms, and added more separation between his torso and his rib cage, and made him his own thing.
But I think when you get to the face, especially, we started with a pretty traditional gorilla. Looking more toward the western lowland gorillas than the silverbacks. Then we just started pushing the anatomy further and further away until we landed at this movie monster that Jordan was looking for, where it has massive brows. He loved the ’33 Kong, how you could read the craziness in the eyes. A lot of that was because that gorilla had such bright eye-whites, and he looked googly-eyed a lot of the time. A huge challenge for us was, “How do we incorporate that look and feel, and still have it look realistic?”
Back to animation
When you have a featured character like this, how do you determine what techniques you’ll use to realize him? Particularly when it comes to performance — do you go through different approaches as to whether to use pure motion-capture, or pure animation?
We definitely did. We were very fortunate to work with [actor] Terry Notary, who I'd worked with before on Warcraft. He did a lot of body performance work. We had a couple days in mo-cap where Jordan could iterate very quickly with Terry to work through different scenes, then also try different gaits. And try things like, “Give us 10 chest pounds.” So he'd try different cadences. Is it three, is it alternating hands, is it hands together? Just trying to give us a nice library of things to pull from.
“Anyone that's six feet tall is going to be able to change direction much faster than Kong would ever be able to.”
Then I would say the same is true of the face. We had a day of capture with Toby Kebbell (A Monster Calls, Warcraft), where he works through some of the scenes — particularly the less action-heavy scenes, where you really have a lot of time to look at Kong's eyes, and the movement of the face. There are some shots where that facial capture is used directly, but through the production process and the reworking of the scenes, a lot of what Kong needed to do changed so much that the capture was used a lot more as inspiration and moments to pull from. And then ultimately a lot of the animation was key-framed. I think that was actually important to do, especially when trying to sell that Kong was 100 feet tall. Because even weighted down and moving slower, anyone that's six feet tall is going to be able to change direction and move much faster than Kong would ever be able to.
It's not even just a matter of saying, “Let’s take that and slow it down by 25 percent.” Once the arm gets moving, it can actually be pretty fast. But then when he needs to change direction, you need to have that appropriate, physically accurate process of getting this massive arm to move a different direction. With the animation in particular, it was a real challenge between making sure Kong felt slow enough where he was huge, but at the same time not letting the shots drag on so long that it no longer became an action movie.
For years, it seemed like mo-cap was touted as this holy grail, particularly for CG characters. It’s fascinating that now we're at a point where you have that tool at the ready, but instead you say, “Well, maybe that doesn’t work well for everything.”
A lot of it was that once we applied the mo-cap onto Kong, the facial capture and the body capture, it worked as a great basis to start from — particularly the facial capture, with some of the subtleties of the eye movement. But in this case, when we originally captured it, it was like, “Maybe he's calm and restful here,” and then by the time the edit came together, when he comes into the scene [we realized] he should be pretty pissed off, and then sort of calm down as he meets Brie and Tom. That necessitated folding in a lot more keyframe animation.
I think motion-capture is incredibly useful, and has gotten really good in terms of the way we can translate it from human to even a creature like Kong. But in this particular case, Jordan also really liked what a lot of the animators were doing with the character. I think that's a great thing. It's definitely worth highlighting the huge contribution they had to the film.
Kong on set
There’s a great moment when Tom Hiddleston and Brie Larson are standing on a cliff, and she just reaches out and touches Kong’s nose. It’s the way you’d approach a skittish horse, and it really cements the audience’s empathy for Kong as a character. What were they dealing with on set for that scene?
It was super low-tech on the show — I think in a good way. They are acting, for lack of a better term. They really had so little there, other than “Here's kind of what [Kong] looks like, here's a thing to put your hand against, and this is about where his eyes are.” They just pulled all the rest of it off from there.
“We used this tool called CineView. It just runs on an iPad.”
When we were on set, we had a couple of things — particularly, we used this tool called CineView, which we've developed here at ILM. It just runs on an iPad. It's a very simple augmented reality app where you can pick proper camera lenses so you get an idea of the right field of view. You can load in [CG] models, and then say, “Okay, this model is 100 feet away from me.” You can see it out in the environment, and say, “Okay, if Kong’s 100 feet away, I need to tilt up this high in order to get to his head.” I think that was useful for the actors, and also for the camera operators.
There usually seems to be one primary technological hurdle every visual-effects team needs to grapple with on a given movie. What was the biggest challenge on this project?
I think most of our technology development was around Kong, and particularly his hair. We were able to leverage off the existing systems we had for muscle and skin simulation, and a lot of the work was just kind of making it work at scale for Kong.
But Kong spends a lot of the movie in the water, or on fire. We wanted to make sure it felt like the hair reacted to that. We never knew when the animators were going to get him wet, and in which way. So we would essentially have animation complete the scene. Let's say Kong reaches down and grabs a squid out of the water. Then we would run the effects simulation to get all the water splashing, and the big surface volume of the water.
That was particularly challenging on the show, because even slowed down, Kong hits the water at 40 or 50 miles an hour. So first take of every effect scene was just a wave that covered him completely. Especially at the end, when he's fighting the Skullcrawler [creature]. That was just displacing massive amounts of water. There was a lot of artful carving of the water to get it to move aside.
Once the water simulation's finished, then we can simulate the hair. [The system] would actually be able to measure how submerged the hair was within the water surface. So if it was fully under the water surface, like at the end, when Kong's rescuing Brie, it would measure the saturation value of the hair and know it was fully saturated. So it needed to change the style of the hair and the looseness of the dynamic, so it really flows in that underwater feel that we typically get. But then as he pulls his arms out, they could dry over time, and then re-clump the way hair naturally would. We spent a lot of time looking at gorillas getting in and out of water to figure out what the methodology needed to be there.
But what was great about the system was that even in later shots where he wasn't necessarily submerged, we had a fully dry version of Kong, and a fully wet version, and then we could simply paint areas of him where we wanted him more wet. It took a long time to get the system set up, but it was incredibly useful. In fact, it also let us, if we needed a bloody area, like around his arm wound, we could paint that as wet, then say, “Okay, this is also bloody, so it's gonna take on a red tint to the fur.” But it would handle everything all the way through shading, darkening the fur, making it shinier, everything to help it feel wet.