Making the second season of ‘Prehistoric Planet’.
For season 1 of Apple TV+’s Prehistoric Planet, befores & afters spoke to directors Andrew R. Jones and Adam Valdez about the challenges of creating a documentary-style series showcasing computer-generated dinosaurs.
Season 2 has now aired, and in this new befores & afters coverage, we talk to MPC visual effects supervisor Elliott Newman to get a sense of the various steps necessary, from a VFX point of view, to realize the most naturalistic scenes possible.
This included planning with concepts and previs, then shooting plates–including with cardboard cut-outs and 3D prints–and then animating creatures in a way that did not feel like a classic Hollywood creature show.
Check out the interview below, which includes a ton of befores and afters images.
b&a: One of the first things I wanted to ask you was how something like Prehistoric Planet differs from a ‘normal’ VFX project for you and MPC, or is it essentially the same?
Elliot Newman: It’s more that the content is different, rather than the tools and process we follow. On a typical film or episodic show, if you’re animating something like a creature, there’s a lot more artistic license allowed. It’s down to the animator to make certain decisions about how something should move. Whereas on this series, we had to base everything around the facts–the science and the records–and what we understand about the locomotion and the anatomy and how a triceratops might walk, say. That was probably the biggest learning curve for us, just having to ground ourselves in the facts and really be driven by the paleontologist, Darren Naish, in terms of producing something that was accurate.
b&a: How did things move from the science to artwork and previs for the show?
Elliot Newman: Jellyfish Pictures worked very closely with Darren on ‘dino packs’. Then, as we at MPC were building each dino, we received these information packs which contained conceptual artwork, skeletals, orthographic views, front, side, top, and any kind of additional information that we needed to inform us about the construction of that dinosaur.
We then handled the previs in-house at MPC in London and LA. The previs was really done in a way to reflect the fact that we were going to eventually go to a location and film empty plates, come back and then add dinos to them later. So, the previs was about how do we make sure that it feels like they are actually recording this action happening in front of us?
We would sometimes recreate these moments as long animation pieces, maybe they were one or two minutes long pieces of behavioral moments. And then in the previs engine we would put cameras, we would figure out our coverage, we’d have an environment that was representing roughly the location we were planning on shooting the type of terrain at, whether it was a forest or a desert scene.
Then our directors and the Natural History Unit from the BBC would come in and say, okay, well how would we document this? If this was something that was happening in real life, what would our logistics be? We can’t get close to a T-Rex, so we’d be further back, right? And you’d be shooting through a forest and there’d be some trees in the way. Ultimately that previs was effectively all rushes that went into the editorial, just the same as a normal documentary. Maybe you wouldn’t have as many hours of footage that you would on a real documentary shoot, but it was kind of in the same vein.
b&a: What were the next steps through to filming?
Elliot Newman: We would cut that down to individual shots and moments. When we had a story that we felt worked, we’d figure out how we’re going to shoot it. What we had was a blueprint. We know what the story is, we know how fast each dinosaur is moving in each shot. We know what the camera’s doing, we know how far away it is, we know what lens it is.
We had these shoot crews that would go to these locations all over the world and we would provide them the previs and we said, look, this is basically what we want you to shoot, and that’s your guide.
As a visual effects shoot, it was, in a way, more of a technical exercise, saying, well, have you captured everything? Have you captured the ref spheres? Have you captured the HDRs? Have we done a LiDAR scan? In that way, it is very different to what a documentary shoot usually is. This visual effects shoot is more methodical, more pre-planned. Every shot costs a lot of money, so you’ve got to make sure that you’re getting everything you need.
Now, given that, there’s always a chance there where you would literally just lose that documentary feel very quickly because of planning everything so much. That’s just how it is to shoot visual effects, and so that’s where the previs really helped to make it feel more authentic because you’re reverse engineering it, you’re putting in the randomness into the actual shot design.
b&a: For the plate shoot, was that a mix of BBC Natural History and MPC crew?
Elliot Newman: It was a collaboration, yes. Rory Bryans for example, was our main on-set supervisor, and he is super experienced in visual effects data capture. Then, partly because of the pandemic, we had quite a lot of shoots happening at the same time around the world. You might usually have main unit, second unit, splinter unit. Whereas here in some cases we had five shoots happening in the same week. We couldn’t have a really big team on each location because we were going to quite remote places and because some of it was during COVID restrictions.
The BBC Natural History team would also be with us. They’d help us in researching locations. They’d be on the shoots with us. There’d be a BBC Natural History director/producer, in a showrunner role, who would be on the shoot as well. They would bring the documentary side of it to the shoots. But if we needed some help LiDARing something, then we’d say to them, well, now you’ve got to LiDAR, too. They’ve never LiDAR’d before! It was all hands on deck.
b&a: Did you still use stuffies or stand-ins or cardboard cut-outs, where it worked, on location shoots?
Elliot Newman: Yes, absolutely. Sometimes the low tech stuff is the best stuff. Cardboard cut-outs were our mainstay. We did have a few situations where we actually did a 3D print of a model. For example, for some of the ammonite sequences, we actually had a 3D model printed because it was an appropriate size to print something. If it was a T-Rex, we would just do a head cut-out that was big enough. For big sauropods, we used poles. The cut-outs are super useful because it just gives you something in-camera. You’ve got something there as reference for the crew to say, okay, well that’s how big they are and this is where they’re sitting.
b&a: What other things on location were you doing to help with integration of your CG dinos?
Elliot Newman: Well, I think the most challenging scenes were the forest scenes where we had sauropods walking through the forests or big predators like Carnotaurus or T-Rex in forest canopies because they’re so big that it’s quite hard to frame them because you’re in a very dense environment with lots of things in the way. So you have to try and figure out where you’re going to shoot from. But also these things would collide into all these trees and branches and canopy. That was probably the hardest thing in terms of shooting, I’d say. Foliage is also quite painful to deal with when it comes to rotoscoping.
So, we used a lot of greens teams as well on locations. For example, we might film at an area intended to be Asia, but we weren’t actually in Asia. So we would use greens teams to bring in the type of plant species that would be accurate, but also in a way that gave us a little bit more of a controlled set. The greens teams would dress things and put things in front of camera. We had the ability then to do bluescreen passes, foreground and background plates, so that we could do clean passes. That was quite useful. So, we’d be on a location, but we’ve augmented the location in a way that was like you would control a complex VFX set on a stage. It at least gives you the ability to pull a key off some foreground foliage and rebuild a background pass that’s clean.
b&a: When you’d done a plate shoot and when an edit was coming together, was there a postvis process, or was it quite easy to translate what you’d already done in previs to the plate work?
Elliot Newman: There wasn’t really a postvis done. I mean it depends on what you call postvis. We had a blocking animation phase, so we would basically put together a plate edit. We’d have the previs edit, we’d shoot all the plates, we then build a plate edit and we’d have a picture in picture of the previs in the plate edit. It was somewhat challenging because a lot of the plates were shot with locked-off cameras. We had high resolution RED cameras and on most of the shoots they would be shooting 8K. That was done for a strategic reason because in documentaries, a lot of the types of cameras are fluid head mounted cameras on a tripod with a long lens, which give you nodal ‘pan and scan’ type camera moves.
We’d do these locked-off plate shoots which would give us these really juicy 8K super detailed locked-off plates to work with. We’d shoot them wider than we intended to final the shots at, which gives us loads of safety. In the animation process we would then punch in and basically build a CG camera move on top. We’d add that ‘pan and scan’ on top. Because we had this great 8K canvas to work with, it allowed us to do that quite easily.
What this meant was that the plate edit was a bit odd to watch. There’d be a lot of locked-off wide shots and then you’d be referring to the previs with all these cool long lens camera moves and it wouldn’t be in the plate. But we knew that was the aim.
b&a: Take me then into the animation phase.
Elliot Newman: There’d be a heavy tracking phase, and then we got into a very early blocking animation phase. At that point the edit’s still in flux. We’d work with decent handles on the shots so we can slip and slide in the edit and make the shots work. You could call that the postvis. As things continue, there are more and more stages of animation.
We might drop shots, or add shots, too. In editorial, they might say, let’s make this a close-up shot now, where it wasn’t a close-up shot in the previs. That kind of thing was easy for us to do. We’d still use the background plate, but it’s more out of focus and we push in on our CG model. We were always flexible.
b&a: When I talked to Andy Jones and Adam Valdez, I remember discussing the idea of animators having to animate naturally and not over-reach on expressions. It was all about keeping it subtle. How much of a challenge was that here?
Elliot Newman: Yeah, I think there’s a muscle memory thing to it. Everyone loves to do a roaring head shaking T-Rex. This not what this show’s about. The direction we were getting from Darren was, well, we don’t think they did that or they didn’t walk that way. He might say, it might not look as good, but that’s what we think they did and that’s more accurate and that’s what we’re trying to do here. So there was, I’d say, a little bit of recalibration for the animation team, especially a lot of the new animators joining the project. I think after a few months you get settled in, you understand the recipe a little bit more, but it would still require a lot of collaboration and liaising with Darren.
It means you can’t really hide behind the beauty of doing all those things that an animator tends to want to do. You have to get the subtleties and the details right, because that’s all you’ve got. The trick is really how do you portray that these things are living, breathing things and there’s something going on upstairs. That is really a skill of just very delicate facial rigging, facial blendshape work, animation studies and just being respectful to the science.
And, yes, there’s a couple of roars and there’s a couple of head shakes. We had to convince Darren on a couple. He hated all of them! But we had to say, oh, come on, you’ve got to give us a couple. Actually, I remember when we were doing our initial triceratops animation studies, they just didn’t walk the way you would think they would walk. So we started out with a walk cycle that was just totally wrong. When we showed Darren, he said, no, that’s not how they were able to walk. The skeletons just wouldn’t allow it, the structure. We know that their joints just didn’t bend like that. They would walk a lot more similar to crocodilians.
It meant that it was harder to animate like that, doing what felt unnatural. These things are massive. They’ve got to traverse these really complex terrains. They’ve got to step over logs and all these things. How are we going to make that work? It wasn’t really about how good our animation was, it was more about how accurately can you portray them in a way that, yes, it’s more challenging to animate or triceratops that way, but we’ve got to do it and we’ve got to do it in a way that looks convincing.
It was the same with the sauropods on season 2. I think there were some fairly recent footprints that were discovered and new information about the gate of a specific sauropod. So we folded that in as well. Again, that was different to how you would perceive the walk cycle to be on a sauropod.
b&a: What you’ve also been successful at I think is the final look and feel in terms of the texturing and rendering and lighting of these dinos. I know it’s about shooting a documentary and not making it with Hollywood lighting, but what was the approach here?
Elliot Newman: In VFX, in lighting and compositing, generally you tend to make life easier for yourself by backlighting everything and putting in a beautiful sky. All these things are done to make something look artistically nice to take the edge off a little bit. But as soon as you’re in an imperfect scenario with plain natural lighting, it’s suddenly all this pressure on the CG to look really convincing because you’ve got nowhere else to go.
For example, ideally if you’re shooting a river crossing, the animals will cross in magic hour. That would be lovely if they could do that, with all the smoke and the backlighting. But that’s not what we can always do, and that makes it harder because you can’t really hide behind anything, as everything’s in broad daylight. Your materials, your shading properties, the texturing, the skin surface details, the way the muscles move, everything’s on show.
I remember looking at the plate edits from a sequence from season 1 fairly early in the production and it was shot in Iceland and it was just the most epic environment. It was just stunning. I knew we had our work cut out for us because we had this incredible environment that we had to put our dinosaurs into and they had to look as good as the environment.