DNEG delves into its sequences on ‘Godzilla x Kong: The New Empire,’ which also included Kong’s tooth extraction, and the enormous crystal pyramid.

DNEG took on several demanding sequences in Adam Wingard’s Godzilla x Kong: The New Empire. Among them was the opening battle between Godzilla and Charybdis in Rome, along with Godzilla’s restful sequences inside the Colosseum.

The studio also delivered a scene of King having his tooth extracted and replaced, and they showcased a biological veil hiding a gigantic crystal pyramid environment, as well as a wealth of creatures inside Hollow Earth.

Here, several members of the team walk befores & afters through the complex creature and environment work.

The insights come from senior visual effects supervisor Paul Franklin, visual effects supervisor Aleks Pejic, visual effects supervisor Lee Sullivan and animation director Spencer Cook.

Godzilla vs Charybdis

Lee Sullivan (visual effects supervisor): There’s the opening fight between Godzilla and Charybdis, the spider-like creature in Rome. It had been previs’d, and they shot plates to match the previs. We handled all of the postvis blocking and figuring out the timing.

Spencer Cook (animation director): Alessandro Ongaro, the client-side VFX supervisor said, ‘We want to make sure that the monsters have the proper scale. And so you guys are in charge of just letting us know how many frames you need for each of these shots to make it look right.’ Previs sometimes focuses mostly on the design of the shot and not necessarily on the realism of the animation. So I think Ale was smart in giving us that kind of latitude.

Lee Sullivan: There’s a big push-in, and that took a while to get right. It was that classic ‘coming at each other’ shot where they collide.

Spencer Cook: That big push-in shot took some time because Godzilla running is a tricky thing to do, when you’ve got a monster that big that can run that fast. We’d never seen Charybdis run either, so we had to design the run of something that big and with spindly legs like that.

Lee Sullivan: And with the plates, I mean, how do you frame for that? How do you even know how tall they are? Imagine the practicalities of shooting helicopter footage over Rome. They probably had some pretty strict limitations on how long they could be in the air. At least if you’re shooting in Paris, you might say, ‘Oh, well, at least roughly half the height of the Eiffel Tower.’

Spencer Cook: Also, we couldn’t use the end of that plate that was shot. We had to cut it off at a certain point because as the pilot got closer, the cameraman tilted down, which would have left us looking at the lower bodies of the creatures.

Lee Sullivan: We had to do a re-time and a repo and we added in plate at the top end of it.

Godzilla tiptoes over the Colosseum

Lee Sullivan: They were okay with us destroying a lot of Rome, but we were specifically told that we could not mess with the Colosseum. Godzilla sleeps in it. It’s like a big doggy bed. We couldn’t destroy any of the existing structures, but we could build extra structures to destroy. We just weren’t allowed to destroy anything that was ‘original’ Colosseum.

Making big creatures do big creature things

Spencer Cook: I learned a lot about animating giant monsters when I was the animation director on Godzilla: King of the Monsters at MPC. It basically boils down to the fact that you need to vary the speed of any kind of creature so that it looks natural. Anything that moves with the same speed is going to look robotic and fake. They can’t move slowly all the time. They can move fast, but the key to that is you have to have a lot of anticipation into the move, the fast move. There needs to be an extra amount of build-up, and extra amount of frames to build up to that speed, that fast speed, a lot of anticipation into the fast speed. And then a lot of follow through after that fast speed to show the dissipation of that energy. You generally use more frames for creatures that are that big as opposed to creatures that would be the size of a person or slightly bigger. But for something that’s hundreds and hundreds of feet tall, the key is extra anticipation and extra follow through.

Lee Sullivan: You’ve also got to worry about the interactivity aspects. For example, there’s a shot of Kong coming out of the water. That anim was signed off on as it is, but then we had to do all sorts of tricks just to get the water to simulate. It doesn’t look like it but those hands are moving almost supersonic speed if you look at the scale and how fast they’re actually moving. So, running a normal water sim doesn’t work–you’ve got to do all sorts of cheated physics or rejig your anim so that it’s three times as long and then run it at that speed and then compress it. You’ve got to do all sorts of weird tricks to get it to actually work for the secondaries.

Spencer Cook: Yes, it doesn’t look like it when you’re looking at the monsters, but for Godzilla’s walk, his normal walk, we put a speedometer on his feet just to see how fast they go. And when his foot is up off the ground and it’s moving forward, it can get up to 200 miles an hour moving forward. But at that scale, it just looks like he’s kind of plodding along in the big heavy pace. That’s another example of the variety of speed that gives it a natural quality and also a heavy quality. Godzilla taking a step, for example, coming off the ground and starting to move forward is slow. And then when it starts to come down, then it speeds up, and that’s where you get that big heavy impact. But the slow up and forward is where you get that variety of the speed and you can keep the sense of weight with that, as well with a hard step.

Lee Sullivan: Sometimes it’s just a question of adjusting gravity. Like when Godzilla’s coming out of the iceberg, we played a lot with how much all of the ice and snow and the volumes of this snow dust, how all of that was falling, made a big difference.

Spencer Cook: I think one thing that helps to sell that, too, especially when the monsters move quickly and we have the big anticipation and the follow through, but there’s a fast motion there and there’s a bunch of debris and stuff, if that debris is moving slow, like either slower than normal or in real world, it moves very slow, then that really helps sell the scale of the motion as well.

Kong gets a new tooth

Spencer Cook: We got all the face shapes of Kong from Wētā FX and integrated those into our pipeline, and then it was just a process of figuring out just how to make that look like the same Kong.

Paul Franklin (senior visual effects supervisor): The biggest challenge there is he isn’t really doing very much. On the one hand you say, ‘Hey, your job’s easy. He’s just lying on his back, breathing.’ But actually, you need movement to sell the character and sell believability. It’s the more still things become, the more challenging it becomes.

Spencer Cook: It was about restraint. Kong is lying there and he’s in pain, but he’s kind of groggy and he’s not feeling well, and then he gets anesthetized. So we had to really dial down the amount of motion. As animators, we want to move everything. But here we had to really tone it down and make it subtle. Especially when Kong was unconscious and we go to those big closeups, we wanted it to not look like he’s dead, so there has to be some motion there. But man, that level of motion, we really had to find just how much or how little he should be moving.

Lee Sullivan: You get some downstream issues from that, too. For example, the first pass of animation was really accentuated by the CFX. And then because we’re so close, that read as too much. So sometimes you have to do a little back and forth again. But there’s some stuff that helps, like a really subtle change, like a rotation on the head that’s really, really tiny, on that scale, you’ll get a little highlight that just moves across the lip or something, and you read that motion that you couldn’t see in the anim version, for example.

Spencer Cook: We were blocking out the animation for all the Kong shots while the asset was still being developed. So at first, we were just doing playblasts out of animation and we didn’t have any fur on it. So we had to get these fur cards so it would look decent in a playblast. And then, once you get that into lighting and put some CFX on it, then it can look different. You see different kinds of motion. In some cases, what looked subtle in a playblast looked like too much when it was lit, so we had to go back in and tone it down even more in some cases.

Paul Franklin: Also, it’s not necessarily the way that a real ape looks. If you actually look at a real gorilla, their skin is quite black, really very, very dark, and quite shiny, and Kong is actually a little bit more humanized. He’s a bit more pink and purple and mauve in his skin tones. There’s a lot of brown, and he’s not quite as shiny as a real African mountain gorilla. And so there’s only so much you can get by looking at real world references. You then have to say, ‘Okay, how does our ape look in this world?’

Aleks Pejic (visual effects supervisor): Kong and Godzilla, I think they both have a similar challenge in that the color palette is fairly compressed. For Kong it’s a lot of brown skin, but we needed to have more range in it.

Paul Franklin: Yes, one of the shots that really pushed us hard was an extreme close-up just of Kong’s eyes and his nose. You don’t see the rest of his face as he’s sleeping. He’s been anesthetized and Trapper (Dan Stevens) is about to land on him. Just the level of detail in the skin that has to be revealed there was huge. We were really working hard to get every last pixel out of the textures that have been applied to it. Obviously we were looking at the work of the other folks, particularly the guys at Scanline. They’d done an amazing job of close-ups of Kong. So that was our target.

The broken rotten tooth had been established earlier in a Scanline sequence, but I don’t think we saw it close up. For our part, we had to see the pulp of the tooth and everything. So there was quite a lot of work done to get something which looked like it was suitably diseased, but also not so disgusting that people wouldn’t want to watch the movie.

Lee Sullivan: We had one part where Trapper, he’s actually standing on the lips. For filming, he’s standing on a static prop, he’s not standing on something soft and squishy, so figuring out how to make all of that feel like it was actually going on was tricky. In the plate we had everything but him standing on a little buck, a little gray buck for the tooth, and that’s the only real thing, and everything else is completely VFX.

Paul Franklin: We also had a lot of work to do with Dan Stevens inside the Mule, which is the kind of heavy-lifting flying vehicle that he flies over to Kong on. Reconciling the lighting on the inside of the Mule to the exterior lighting–there was a lot of work done there, balancing that, because it’s quite diffused and very orange inside the Mule, so we had to justify the light coming in, reflecting off the paint, and then reconcile that with the fact we then tilt down, we see Kong asleep below. So we did a lot of little ‘iris pull’ tweaking in the sequence, basically changing the exposure over the duration of a single shot. There’s quite a lot of complex digital cinematography work going on. Very subtle. People hopefully will never notice it, but it’s difficult getting those things to work.

Motion capturing for Kong

Spencer Cook: We did mocap for Kong, and I set up this remote session, which was really interesting because it’s the first time I’ve ever done that remotely. I’ve been on many mocap stages and directed mocap for many projects in the past, but always being on the stage. DNEG has this really great mocap stage with this really big volume in London. We used that. I was in Montreal on a zoom call. They set up cameras in there so I could see everything. They had a big monitor, and so the performer could see me and I could talk to the guy. I made a list of everything, every sequence that we were doing with Kong. Ace Ruele was the performer and he did a great job. It gave us a very good foundation.

Of course, we ended up keyframing a lot of Kong stuff. How movies are made today is a very iterative process. What they originally shoot, and what they originally think they want often changes. I shot all of the stuff with the information that I had, and then as the shot progressed, Adam asked us to do different things with Kong. So the mocap started to fade a little more into the background and the keyframe took over for most of the Kong shots.

Godzilla arises from the ice in dramatic fashion

Spencer Cook: There are similar shots in almost every Godzilla movie where he’s rising up and we see his back fins and then his head comes up. So, the main thing for me when doing shots like that is making it look like Godzilla is actually planting his feet, planting his hands, pushing himself up, and not just rising on an elevator.

Lee Sullivan: We really worked hard on the FX for that. Realistically, if you have something the scale of even just Godzilla’s back bursting out of the ice, it would create an avalanche and a cloud of snow dust that you wouldn’t see anything. So a lot of it was managing our sims aesthetically to try and not just channel the layers of simulated effects, but to make sure that it’s broken out into segments that then you can art direct it.

‘Fly Guy’ and the giant mosquito

Spencer Cook: There’s a dragonfly that the character Jia sees at the temple.

Lee Sullivan:I just called him Fly Guy.

Spencer Cook: The key with insects is that they have a lot of fast twitchy kinds of motion. When he walks, he kind of scuttles and stops and pauses. So all of those characteristics are insect behavior. Quick, sharp motion that have hard stops to them. The wings flutter in a really quick way. The key story point in that sequence is that the Fly Guy is actively consciously leading Jia to that rock lever that she pushes in. When he flies around, it’s just zipping around and there’s not a lot of detail on where he is looking and stuff. But when he lands, we were consciously choreographing it in a specific way so that he would look in one direction and then kind of move that way. Then when Jia’s hand starts to approach, he looks over at her hand. By using the direction of his head, it gave it a little bit more of a conscious body language, that he’s there to specifically say, ‘Look, push this thing’, not just like he just is an insect that just randomly landed there.

Lee Sullivan: Then we also had the mosquito that lands on Trapper’s arm. In terms of the look of those, that was a big challenge. Especially Fly Guy, because he does have a bit of a cartoony aesthetic to him. You can show that in a turntable, and it’s a very bright cuddly, fuzzy creature. And then when we put it in this really dim, mysterious, intimidating temple environment that they’re in where they see him, it was actually quite challenging to make its cute fur read. You’re dimming it down to try to make it look like it’s in this space and suddenly it doesn’t look that cuddly. He looks kind of weird and mottled. So there was a lot of rebalancing.

You would think the mosquito was the most straightforward because it is the most like an actual physical mosquito that we’ve seen before, but of course it’s in Hollow Earth, so it’s bigger and the proportions are skewed, and it also has a bit of a sheen to it.

Mikael is eaten by…a tree

Lee Sullivan: On set they had a static tree trunk prop and the actor would pantomime getting grabbed from behind. We then did a digi-double takeover. He actually completely folds up and gets sucked inside.

Spencer Cook: You can’t imagine how many versions of that we went through. Adam had a very specific idea in mind for that, but visualizing that in so few frames was surprisingly difficult.

Lee Sullivan: We initially did a version where there are tentacles that grab him, and it was all very woody, but it wasn’t reading that well. So we ended up skewing the tentacles, making them a lot more slimy and pink as well, to make them show up in the dim environment.

Behind the veil

Paul Franklin: The characters push through this bionic or biological barrier before they discover the pyramid. I likened that to the sorts of things we used to do on the Harry Potter films, in that it’s essentially a magical effect, but you want to sell it as something that’s actually happening there on the set in front of the camera. It had a lot of contradictory characteristics. The veil is supposed to be this energy barrier and it’s got an organic quality to it. It was not entirely clear how this thing has been created, but it appears to have been grown, since it had a fibrous veined structure to it. It’s also reflective, so that when you are approaching it, you don’t notice it to begin with because it’s reflecting the world around you. So it’s essentially like a giant mirrored Mylar sheet, but not a perfect mirror.

Interestingly, the visual effects editor on the production side, he had just dropped a placeholder effect in that he had created, possibly in After Effects, where he’d taken the plate and manipulated it. And of course we ran into that thing that happens, which is when a filmmaker is working with something and it’s been in there for such a long time, they’ll either end up hating it or loving it. There’s no sort of in-between, really. And in this case, they loved it. So we were quite far into the development of the veil, and Alessandro then presented us with the work that the editor had done and said, ‘Well, the director really likes this. He really likes the color palette, so we want to try and incorporate aspects of this.’ That took us in a different direction. The final version of the veil is the synthesis of lots of different ideas.

Aleks Pejic: We ended up creating a series of layers in Houdini for the veil. It was tricky to match the concept art and then go into shot production and keep that concept art look. We’d get to the point of it being 90% close enough, but it didn’t quite look the same as the concept. So we really had to balance so many things for it to work.

Paul Franklin: The reflections of the cast were all created with body track digi-doubles, which were produced to a very high standard of detail, because we couldn’t use the plates. There were no reflection plates to be used there. On set they had a fibrous sheet that they could interact with and press against and tear apart, which gave us something to work with, but they wanted it to feel thicker and heavier and more resistant. All those fibrous edges as they tear it apart, that’s all created in Houdini. There’s nothing of the original practical element left. All the atmosphere that comes through, the little particles AND moats that are floating around, we added that all in after the fact as well.

Inside the crystal pyramid environment

Paul Franklin: That environment was a big creative challenge. There were various requirements for the pyramid. First off, it’s a pyramid. It’s really enormous. It’s bigger than the Great Pyramid of Cheops, outside of Cairo. So it’s absolutely huge. It’s made out of translucent, glowing crystal with some sort of mystical properties. And then it’s very ancient, and it’s covered with these hieroglyphs. So we had to work out, what does a giant pyramid, hundreds of feet tall, look like, that’s glowing, that has all these properties to it?

The reality of real crystal, if you look at things like amethyst or quartz, which was one of our key references, they’re too opaque. They cloud up very, very quickly. If you made a large block of quartz, the light would scatter in maybe a few inches into the surface. If it was amethyst, which is a type of quartz which is clearer, it would scatter a little bit further, but then you wouldn’t be able to see very much. But we wanted to see into the depths of this thing. Ours needed to be clear. It’s clearer than an iceberg, for example. It took quite a long time to figure out the creative balance of those ideas. We did a lot of work with our own internal art department. Aleks Pejic and I spent a lot of time searching for references on the web, looking at all sorts of things, looking at crystal ornaments in my home, even. We ultimately produced something which everybody liked, and then that could give a target for our shader writing team to develop the look.

It’s such an abstract object. How do you sell the scale of such a thing? If you think about the real pyramids, if you go to Egypt and look at the pyramids, one of the things that gives you a clue to its scale is that of course it’s got a stepped edge. Originally, those pyramids were covered in alabaster or something, and over thousands of years, most of that material was taken off and used to build modern Cairo. So you’re just left with the big sandstone blocks on the inside. That stepped edge gives you a scale reference. We didn’t want to do that with our pyramid. We have the idea that it’s aged, it’s broken, it’s got little fragment details, but essentially, they’re fairly straight lines. So there was a lot of work done with atmospheric perspective and adding little depth cues like the clouds that we see floating above it–all those sorts of things which give you an idea of how big this thing might actually be.

Aleks Pejic: Yes, one of the tough things was dealing with something this big for different kinds of shots–wides and then also close-up shots. We were worried that it was going to take 16 days of rendering just for one iteration.

Paul Franklin: So we ended up taking different approaches for the different types of shots. The super-wide shots are rendered in 3D, but essentially there’s quite a bit of digital matte painting and then very, very careful compositing to balance it all in with the rest of the environment. The same is true of the very large glowing green crystals that we see in the Hollow Earth environment that surrounds the pyramid.

And then there’s Jia’s ascent of the staircase when she goes to the very top of the staircase that runs up the face of the pyramid. That required real attention to how the light was refracting through all the close-up details. We see her feet on the steps, and so she has to actually bed into that thing. There was a lot of interactive lighting on the set, but it wasn’t necessarily motivated by the kind of pyramid that we were going to ultimately create. When they shot that, they didn’t know what the final look of the pyramid was going to be. So, again, there was a lot of intervention in the elements to bed them in with the digital pyramid.