In Land of the Lost, a Brad Silberling film starring Will Ferrell as an alternate world explorer, Rhythm & Hues created hundreds of visual effects shots to depict a universe populated by dinosaurs and other creatures. Ian Failes talks to visual effects supervisor Bill Westenhofer about the project.

fxg: Which shots did Rhythm & Hues do for the film?

BW: We did the vast majority of the effects – pretty much everything except for the stuff inside the pylon, the white crystal table. Those shots were done by Hammerhead. Another company did some footprint removal for the sand dunes at the beginning. We did the dinosaurs and other beasts, the set extensions, the waterfall and raft sequence at the beginning, the Zarn creature and the Sleestaks at the end.

fxg: Can you talk about your on-set role?

BW: As the overall vfx supervisor, I was involved from the beginning in breaking down the script. On set I worked with Brad Silberling in working out how to shoot the elements and figure out things like the choreography for the dinosaurs. Myself and my animation director traded off roles being the T-Rex, Grumpy, which on set was a 15 foot windsurf pole with a tennis ball at the end. So we did the best we could making menacing sounds to scare Will Ferrell and give him that motivation as he’s being chased.

fxg: I’d like to talk about the dinosaur shots, especially for Grumpy. How did you plan those shots?

BW: I might start with the most difficult dinosaur shots, when Will was riding Grumpy, because that encompasses everything. With something like that, we started with the storyboards and then we jumped immediately into previs. So before we had shot anything, we had measurements of the set that was going to be built and we blocked out the various shots with Brad and determined the choreography for what the basic events were going to be.

Then we took that previs on set and shot the plates first. So for the shots of Will riding on Grumpy’s back, we knew from the previs how fast the T-Rex was moving and how much space it would take up. We used our 15 foot pole to block off the action. 15 feet proved to be a perfect number. If you held it at the right angle, say slightly forward, it would give you a good spatial reference for where the T-Rex’s nose would be. The dinosaur was about 13 feet tall at the hips. With another few feet of Will’s torso, 15 feet was just the right height to give enough room for both of them in frame.

We went through and shot the plates completely blank, no T-Rex, no Will, no Sleestaks or anything, just to get the camera moves right. We then had four weeks to take all the plates that were shot, bring them into our system, track them using our software and do the animation with the dinosaur and Will and get it to as close as final animation as we could. That’s kind of unusual. You don’t usually get to that level of animation while the shoot’s still going on. But we needed to, because we wanted to take that final animation and send it through a bunch of complicated maths routines to take the motion of the T-Rex and the motion of the camera, and combine it all into another motion base, that had a saddle shaped like the T-Rex’s back, and a motion-control camera. So we could put Will Ferrel on top of that base and have him move through the frame at the right height and right speed as our animated T-Rex was moving.

The very last thing that we did, then, was shoot Will against bluescreen on the saddle. It was actually really great that we were able to use Will for real. A lot of times people are using digital doubles, but this is a comedy and Will is so great at ad-lib that we’re able to get his performance, even though he’s sitting on the back of a T-Rex. A great example was when he came barging into the scene. We’d animated some action of Grumpy reaching over and grabbing one of the Sleestaks and tossing him away. Before Will got up on the saddle, he saw what we’d animated for the shot, and when he got up there he said: “Where is that Sleestak?” He didn’t really tell us why, but he asked where it was, and we figured out where it was and gave him an eye-pole. So the shot starts, and imagine Will sitting on top of the T-Rex and coming into the throng of 1000 Sleestaks, and he goes: “Fight that guy, right there!” as the T-Rex goes right over in time and bites the Sleestak and tosses him. That was just something Will came up with on the fly.

Another example was when he was running around in the desert and the T-Rex picks him up by his backpack. It was sort of an homage to The In-laws, where Alan Arkin did this serpentine thing to avoid getting shot. Will says that T-Rexes don’t have rapid reflexes so if you run a serpentine pattern you can get away. He was actually getting picked up on location with a flying harness and yanked in the air. Before we shot it, Will said he was going to try and keep saying “serpentine” as long as he could. And sure enough, he gets lifted up by the harness and he’s really getting thrown around up there, but he does his best to shout out “serpentine”, which is perfect as the T-Rex picks him up and shakes him back and forth. So it was fun to get little things like that on set.

09Jun/lotl/2368_GC035_fxg: Can you tell me about how you designed and built the dinosaurs in the film?

BW: Well, even though it’s a comedy, the T-Rex is as state-of-the-art as we could possibly make it. We think it’s pushed even beyond the Jurassic Park-type dinosaurs in that all the muscles are firing right and the skin’s sliding right and wrinkling properly. We did as much procedurally as possible, but there’s a lot of artistry in there in terms of the technical animators firing those muscles. The animators might kill me for telling you this, but in order to get muscle reference they actually put some baby oil on themselves and shot themselves walking around as T-Rexes to see when the muscles were firing. There was a lot of dedication from them to get the details right.

We probably could have gone without those details for this show, but we really wanted to push it and get the T-Rex to be as realistic as we could. We took a few artistic licenses, but for the most part it is a real T-Rex. We added some scales on the back of its neck, for example. A paleontologist who we’ve talked to a number of times was quite happy with the work. One of the reasons to make him as real as possible was so that when Will points out that he’s got a wallnut-sized brain, he finally breaks into the traditional charging beast to having a personal vendetta against Will for the rest of the film and it makes him more powerful if you thought he was a real traditional animal up to that point.

09Jun/lotl/2368_FS045fxg: What approach did you take to modelling and animating the dinosaur?

BW: We use Maya for a lot of our modelling here. We also used zBrush to do the detailed work. Our art director, Chris Grun, is also an amazing creature designer. He guided the zBrush artists to put in all the scale detail. We’ve done films in the past with other animals where you gather up as much reference as possible. With the T-Rex, there aren’t that many around to get reference from, so we go to every other thing that we can. We went to the zoo to look at elephants. We patterned a lot of his surface scales on tortoise scales – just their reflectance quality and the way the dirt gathers between the pores. We found nature where we could to base our work on.

Our animation is done in Voodoo, which is our own proprietary package. Erik De Boer, our animation director, led a team of about 60 animators. Two-thirds of them were in Los Angeles and the rest were in Mumbai and Hyderabad in India. We did our animation blocking and key-framing and then that was handed off to our technical animators to do all the muscle and skin work on top of that.

09Jun/lotl/2368_MK003fxg: There must be a large environment half to your work on the film. What work was involved there?

BW: Well we did about 536 shots in the film, 200 were dinosaur shots, but the vast majority were environments. We were always going to be doing a lot of environment work, but that grew exponentially on set. Bo Welch made these beautiful sets that the actors could act in, but part of what I’m learning is that comedy tends to get shot in wide-angle lenses. So matter how big they made these sets, we were always seeing off the edge, or another angle was funnier so we’d see into the silks and other equipment. It was a big challenge because there wasn’t always a lot of opportunities to put bluescreens in where I wanted to. Our roto artists did a tremendous job of getting around that.

Our environments included the home cave, the two sets where the Sleestaks occur, the dumping ground, the crevice and the vine swinging area. We achieved them using pretty much every trick in the book. Some places for where it’s just a little piece we’d do a traditional matte painting. For others, we used an internal package called Rampage that allows us to take anywhere from high detail to low detail geometry and project matte painting textures on that. Even with a moving camera, you could paint something, scroll forward and reveal which area got revealed by the motion and paint in a patch that way. Finally, we’d do completely 3D extensions. The Sleestak Temple set was only finished about half way up the walls, so we built the top part of that and added digital trees and the mountains in the background. The 3D extensions of course let you get the right parallax movement.

09Jun/lotl/2368_ST091_fxg: What was involved in the earlier cave and raft sequence?

BW: There was a physical set built, which covered the first part of the sequence of the raft going along in the water. We did add some debris falling from the ceiling which would have been too dangerous to put near the actors. At the end where the cave starts to break apart and they fall into the abyss, that was done with the actors on a motion base moving around with some interactive lighting. We added the cave, the water, the debris and the wormhole they get sucked into. For the final tip over the waterfall, they were on a motion base, and as the camera pulls away we take over with a CG double and CG raft falling into the light at the bottom.

We used our fluid simulation software called Ahab. We received a technical Academy Award for that last year. We had to push it a little bit farther to match what we were seeing on set. It let us add an even higher resolution for the ripples, the foam that’s sloshing around, and to get the underwater jacuzzi lights they had established on set to work right as well. The vortex was done using a particle system. Rhythm & Hues uses mostly proprietary software, but here we used Houdini for the effects work.

09Jun/lotl/2368_MB022fxg: What were some of the other creatures you worked on?

BW: I love the shot of the Allosaurus getting frozen. The explosion for that was all digitally created. I also love the compys. Those are the ‘DVD’ characters. If you sit and scroll through the DVD, you’ll see some great detail. Typically in a scene there were 25 to 30 of them, so they’re all doing something funny. I encourage people when they get the DVD to scroll slowly when the ice-cream truck first arrives, and they’ll see the couple of unfortunate compys who are hopping around on the raft and realise just before it’s too late that they’re about to be squished.

The giant crab scene was fun too. Brad wanted to be a little ambiguous at first about the scale of the crab. It’s part of their hallucinogenic trip and you’re not sure at first if it’s a tiny crab running up. So even though it was massive, we tried to keep the animation on a scale that left you in question as to how big it was. It’s not until it rears up and you get the wide angle that you realise it’s pretty huge.

We also did the pterodactyls and the babies for the chorus line sequence. That was a set piece that involved a set extension. There was an attempt made to do more of that practically using silks and mist for the clouds. Part of the problem was, in order to get that smoked up enough, they had a hard time seeing on set. So we ended up adding all of the clouds after the fact and building up that environment. All the cracking of the eggs was done digitally. The goop was created in Houdini. We did animation cycles for a lot of the babies and hero animation for the guys in the foreground. The shells were set up so that they would crack appropriately based on where the guy was bursting out.

09Jun/lotl/2368_F0136Rfxg:: What digital work did you do for the Sleestaks?

BW: The Sleestaks were intentionally men in suits. Brad wanted them to be an homage to the original series as much as possible. For the most part, when you see them in the foreground, they are men in suits. Our role in the film was to add in background characters. We used crowd software Massive. We went to a motion-capture stage and captured a bunch of actions. Then we wrote programs in Massive which are essentially the brains of these characters. Based on the environment and what’s happening to them, the brains make them decide what action to do. So you can get this almost emergent behavior of digital characters. When Grumpy comes in and starts crunching everybody, all the Sleestaks were digital so that we could have them react appropriately. We’d use motion-capture wherever possible.

Another character we worked on was The Zarn. Originally he was going to be more of a particle effect. In the original series, The Zarn was basically a guy in a black suit with Christmas lights stuck on him that were double exposed in the film. So at first we were going to try something that was a homage to that. We had a particle effect for that, but it was kind of confusing to people who weren’t familiar with the series what The Zarn was, and to explain the role of Enik, too. So late in the game, they decided to make The Zarn a real creature and modeled it after the Altrusions, which is what the Sleestak race is.

In a month, we had to re-purpose a digital Sleestak model to make the newest Altrusion. We ran back to the motion-capture stage and did some rapid some new capture for The Zarn’s performance. Then we added the video effect and put the whole thing together. All those things were done in about four weeks from start to finish.

fxg: How big was the team for the film?

BW: We had about 400 people working pretty heavily from November 2008 to May 2009. There was a smaller team including myself on the project for a year and a half.

fxg: I think for Land of the Lost, the visual effects seem just as much a part of the process and a part of the story as anything else.

BW: Our goal is that we want to be part of the process. It started with getting away from the former necessity that you had to do lock-off shots for visual effects. You want visual effects to be part of the normal shots and it has to flow. With Grumpy, even if dinosaurs really existed, you’d probably realise that it was an added digital character, but your goal is to get it to that point where the mind forgets and you just absorb it in with the story.