hardest things

The Hardest Effect I Ever Pulled Off

42 filmmakers, cinematographers, and effects artists on making the impossible happen.

No New York Cities were harmed in the faking of Deep Impact’s giant wave Photo-Illustration: Vulture/ Photo by Paramount Pictures
No New York Cities were harmed in the faking of Deep Impact’s giant wave Photo-Illustration: Vulture/ Photo by Paramount Pictures
No New York Cities were harmed in the faking of Deep Impact’s giant wave Photo-Illustration: Vulture/ Photo by Paramount Pictures

For nearly as long as we’ve had motion pictures, we’ve had illusions to put in them — and they’ve never come easy. The year 1893 saw the first public viewing of a kinetoscope film, Blacksmith Scene, and just two years later, early filmmaker Alfred Clark executed what is commonly regarded as the first special effect. In a gruesome short about the beheading of Mary, Queen of Scots, Clark stopped the film just as the blade was about to chop off Mary’s head, instructed the actors to freeze, took the one playing Mary off the set, replaced her with a dummy, and started the film again before bringing the poor dummy to its decapitated fate. Around the same time, the pioneering Georges Méliès developed a similar switcheroo, then went on to even mix live-action and animation in 1902’s legendary Le Voyage dans la Lune. The race to innovate hasn’t stopped since, and it has always involved pushing the medium to its limits.

As part of Vulture’s weeklong series of stories about the wonder of special effects, we spoke to 35 filmmakers — directors, cinematographers, effects artists — about the toughest effect they’ve ever pulled off. The resulting stories run the gamut from the computer-generated to the practical, the spectacular to the subtle, and all of them remind us of the sweat that goes into making movie magic.

This story was original published in 2018, but has been updated to include more recent conversations with VFX supervisors. If you’re a VFX artist or SFX designer who has a story to share, let us know at [email protected].

Game of Thrones, seasons six and seven (2016-2017), Daenerys riding a dragon

Derek Spears, VFX supervisor, Pixomondo: In season six and season seven of Game of Thrones, we started putting Daenerys on top of the dragon. One of the things we learned, doing season five, was that when those dragons flew out of the arena, it looked a little bit too simple, because the cameras didn’t move. We thought, well, next season we should move the cameras more, and move her more, too.

The production came up with a system of using moving cameras and putting Emilia Clarke on top of what’s called a buck, which is a wooden form for the dragon. That would be moving, everything would be moving, and we planned out all these moves ahead of time, planned in motion control. We thought, okay, this will all line up, and it will be perfect, because we did so much planning. It’ll look great.

What we found out is that dragons aren’t like solid pieces of wood. They move, and when they move, they flex, and their muscles move too. So we had the dragon flying and the muscles moving — but Dany’s legs didn’t react to any of it. So we had to replace her legs. We ended up replacing her legs on so many of the shots just to get it to follow the trap muscles on the back of the dragon. It was a nightmare. Everything else in those sequences was easy compared to trying to get her to sit on top of the stupid dragon with its stupid muscles. Some of the shots, we replaced her legs completely; others we mapped them on to a new piece of geometry and moved them around. Literally we had to deform the lower half of her body to make it fit.

It’s one of those things: You look at the dragon sweeping over the ocean and there’s fire and all these amazing things happening, and the hardest thing in the world is having her stupid knees not slide on top of the dragon. That one little detail was probably half the labor that went into those shots. Just getting her to sit down. The first time we did it, we had no idea how hard it would be. The second time, it didn’t get any easier. It was just as hard.

Rogue One: A Star Wars Story (2016), CGI Peter Cushing

John Knoll, effects supervisor: Peter Cushing had of course passed away years before, meaning this wasn’t a situation where we had the subject available for consulting. So we found an actor to play the character [Grand Moff Tarkin] on set. We’d replace his head or his whole body with our generated character afterward, but we wanted someone in the space, giving the other actors someone to play against. I think we did really well with Guy Henry. The one thing that he said up front was, “I’m not a mimic. I can’t do an impression. But what I can do is play the character with the same dignified, crisp speaking tone.â€

We photographed him on set, wearing a head-mounted camera system so we could track his facial features as he moved. We built a CG model of Peter Cushing using footage from the first trilogy as our reference point, and used Guy’s motions to animate the CG model. I was maybe too optimistic in the beginning, but we found that a great deal of likeness comes from the muscles of your cheeks and lips and mouth. When Peter Cushing made an “ah†sound, he mostly moved his lower lip and exposed his lower row of teeth. Guy didn’t really do that. When he said the lines that had that sound in the dialogue, it wouldn’t look like Peter Cushing anymore. That was one of the big challenges, improving the likeness in little details like that.

This kind of work is extremely sensitive, because there’s always a point early on when it looks off-putting and terrible. It’s not until you’re about 98 or 99 percent done that it starts to look not-horrific. The uncanny valley is definitely a pitfall.

A Girl Walks Home Alone at Night (2014), pile of dead bodies

Ana Lily Amirpour, director: In A Girl Walks Home Alone at Night, there was this ditch of dead bodies, and we didn’t have the money to get, like, fuckin’ 200 extras to go lay in a ditch. I was in Taft, California, and I shot the ditch, and then my producer actually took off his clothes and laid in the ditch in a few different spots so we had a few different angles and positions of him. Then back in L.A., we photographed actual people laying down in different positions. So all of the bodies are actual people, but then that was made into a matte painting by my visual effects artist. We used the power of compositing in CG, but it’s real people that are being composited together … so that’s why it looks real. We wanted to get the real motion of somebody dragging a body and rolling [it] into a ditch, so we did that for real — we put a dummy at the bottom of the ditch so that when he actually rolled, he hit something, and you’d see that point of contact. All the other bodies are a matte painting that got added to it. Computer effects are so much easier, no one bothers doing anything real anymore. It’s all just on a green screen and everything is made like in a video game. When Richard Donner made Superman and made all those miniatures of New York City, it was so beautiful … and expensive. It involved hundreds of artists, building that set. It’s so awesome​-looking, though.

A Quiet Place (2018), cornfield ramp/final creature

John Krasinski, director: The idea of these kids drowning in this corn was something that I was terrified of the entire process. I remember I had spoken to a friend of mine, Drew Goddard, and his only note at the end of the script was: “How the hell you gonna do that corn scene?â€

How we did it was Jeff Beecroft, my amazing production designer, designed this ramp under a small kiddie pool of corn. So there were only 12 inches of corn the entire time for these kids to look like they were drowning in. And they could walk up and down the ramp at any time, and look like they were above the corn or below the corn. If they ever got nervous, they could just duck below the corn and gasp for air. But it was an intense day. The kids, obviously, were such troupers and it was one of those things where [they’re] drowning in corn for, you know, eight hours straight. But to this day it’s one of my favorite effects in the movie.

The hardest visual effect would probably be at the end of the movie — the whole showdown with Emily [Blunt] and Millie [Simmonds], when Emily is pointing the gun. Trying to create the creature in a very confined, small space, to design where he could go and where he was coming from, and how he would attack them at the very end, was very hard. The fact that his face opens up and each one of these little plates is a different sort of radar device was phenomenal — and really, really difficult to nail. We weren’t even sure we were going to get all the effects into the final movie because we were so crunched for time.

I ended up playing the creature — I put on the mo-cap suit, which was Scott’s [Farrar] idea, which I loved it. So I’m the one standing there in a way-too-tight suit, making little roaring sounds for the actors. But it was all Scott and Industrial Light & Magic putting the creature in, and figuring out how he would move through the room to get to the final gunshot.
They basically painted the creature over me.

Maniac, car crash (episode two)

Ilia Mokhtareizadeh, effects supervisor: The car crash when Emma Stone and Julia Garner are on their road trip was really tricky. Because it’s a car crash scene with A-list celebrities, we couldn’t really do anything practical with them. So their entire scene was shot on green screen, which is not unusual. What was unusual was the lengths we went to for the actual crash. I just drove around in a Porsche Cayenne with a Russian Arm on it, to get every single angle we possibly could. And then, using special effects, we had a bunch of witness cameras on the crash cars and the truck. Basically we just crashed everything together. And unfortunately, it all went wrong. Some cameras failed, some got destroyed. It … wasn’t a good day. Let’s put it that way. So we had to think on our feet, shuffle back to the stage, and rethink some of these pieces. A lot of the crash cars we ended up having to create from nothing. Some of the really intense pieces we created using photogrammetry, where you take stills of your environments or objects from every angle to build a 3-D object or environment out of that.

We ended up having to repurpose some of the practical crash stuff. The shot where our Jeep is coming straight toward the truck, just as we’re about to swerve out of the way — that shot did not exist. What that originally was was a live-action Jeep being dragged across the road by a big old clunky rig. That whole thing failed miserably. So what we did was painted out the entire car, rebuilt the whole environment and got rid of all the rigging, and put a CG Jeep in there. Which is kind of scary, because you have a whole piece that’s physical with this one CG car. The human eye is very good at spotting the difference between a CG object and a real one.

In the shots where the car is spinning in the air, and we’re in the interior, we composited everything realistically. The background was spinning much faster. But Cary [Joji Fukunaga, director] noticed that in one of my early tests, I just slapped a background in that was not the correct speed. Much slower. And weirdly, it gave it a dreamlike effect: the sky, ground, sky, ground out of the windshield. Because it was moving quite slowly, it was this happy accident. It felt much more cool and interesting with the incorrect speed outside the window. It speaks to, when you’re in a very traumatic event, they say time slows down. Usually, I do what is technically and physically correct. But this one time, we thought, “This is a good little mistake we made.†Watching the completed scene, I was relieved, because nothing went according to plan on that scene. I feel like we salvaged it really well.

Deep Impact (1998), water coming across New York City

Mimi Leder, director: The hardest visual effect I’ve ever encountered was when I was making Deep Impact with ILM, creating the water coming across New York City. It was the very early days of water technology, so to make that authentic and real was something, at the time, that they were working very hard on, and I was, as well. To make that feel real and authentic and feel like you were in it. The speed of the water. The look of the water. The whole thing. The waves shooting up on the beach with Téa Leoni and Maximilian Schell, and her saying goodbye to her father. Making sure that the effect was so real that it didn’t take you away from the real emotion in the film. They knew that they were going to die together. And it was very important that the emotionality of the scene remain pure and honest, and that you stayed focused on that while this humongous wave was coming at them. So Scott Farrar, he’s a genius — he had to work with very difficult shots — I was doing crane shots really high, coming down, zooming down to where they were. I was always moving the camera. There was only one shot, and it was that shot, the last shot in the film, where the water recedes, and the rotoscoping is not perfect. You could tell it was off. And not because they didn’t have enough time to do it, but because the technology wasn’t advanced enough to do it.

Apollo 13 (1995), launch sequence

Robert Legato, effects supervisor: In the previs phase, which is “pre-visualization,†we would use the miniature models to build out the storyboard that we later break apart into pieces that comprise the sequences. These days, I can use a VR tool to put together a storyboard that lets me visualize the scene on a more accurate scale. However, back when doing Apollo 13, we had to literally build a miniature mock-up set shaped out of foam core to visualize the set in previs. So instead of building a 400-foot rocket model, we did a 40-foot version that’s more practical to imagine scenes around.

We then would go up to the foam set, in this case a rocket ship to resemble a NASA location, and use what we called a small “lipstick†camera the size of a magic marker to hover over the subject. That camera can get in and around the miniature rocket to capture what the scene will look like. Once we got this low-quality footage, we got a better sense of the scene’s tone and could make decisions about potential wider shots, close-ups, quicker camera moves, or whatnot. For example, we would add a shaking camera in the style of a real spaceship launch to breathe life into the footage. For things like the Apollo “rocket smoke,†we literally used a fire extinguisher to replicate what the launch looks like and fool the eye. So we used close-up shots of it “launching†into the same direction in order to mimic that grand effect of the real-life scene. To the viewer’s eye, it looks like we built a rocket ship, which is obviously too expensive and impractical for a scene. Not to mention, dangerous!

American Gods (2017), man-eating vagina

Bryan Fuller, showrunner: The first thing that comes to mind is the man-eating vagina sequence from American Gods, because that was something where we had so many different moving pieces. We had a woman —who was getting bigger with every shot, and a man who was getting smaller with every shot, in a room that was staying the same, and a bed that was staying the same. Then, we had to put the shrinking man and the growing woman in the same shot, and have him being pulled into her vagina. We used a variety of different effects to try to tackle it, but even looking at the sequence now, I look at it and think that some images feel compressed. There’s one shot in particular where I’m like, “The man looks like he’s compressed to a 2-D element, and he’s supposed to be 3-D going into a 3-D element of a woman who is growing larger,†so from the shifts in movement to track between her thighs and his stomach and the skin textures — it was a black woman and a white man — there were so many things that were combating for authenticity and reality. It almost got there! But we didn’t quite pull it off, and I suppose those are the more interesting stories, when you don’t quite nail it.

With a woman swallowing a man with her vagina, you would think like, okay, well, you need a prosthetic labia or turtleneck to pull over the shoulders and neck of the gentleman being absorbed. That was something we looked at — we looked at a big latex labia, and it just seemed silly! It was the strangest shawl collar that you will ever see on the runway. It didn’t quite secure the surrealism of the sequence that we needed because it was physical and it was right there on an actor. You could see that the skin texture of a giant vagina was different than the skin texture of a normal-sized man who was shrinking to get inside of a goddess. We didn’t want to give the audience anything to giggle at, and that was part of the difficulty. Fortunately, Yetide Badaki, who plays the goddess, was incredibly comfortable with her body and her sexuality, and saw this as a wonderful opportunity to be able to show women everywhere that they are their own goddesses, and they are in control of their sexuality. So, in a way, the best visual effect in that sequence was Yetide Badaki’s performance.

The vocabulary in dealing with a highly sexualized visual effects sequence is also always really, really entertaining because everybody there is a bit of a nerd and a professional who is used to talking about mathematics and X-Y axis and how things can feel real even though they are completely unreal. Whenever you bring in a sexual component, it becomes very amusing.

John Wick: Chapter 3 — Parabellum (2019), the knife scene

Rob Nederhorst, VFX supervisor: On set, we have what we call “rubbers†— rubberized knives. The prop department takes a real knife, casts it in rubber, and make a mold of it. Those things are indistinguishable from the real thing. They’re phenomenal. The problem is, if you take one of those and slash at someone, if you go fast enough, you’re going to cut them. And we don’t want that. Safety on set is the number one priority for everyone. You cannot stab someone or slash their throat with a rubberized knife.

Instead, we make a handle, so the performer has something to grip on to. And then we match-move them on a computer and put a knife in there digitally. Ninety-five percent of the knives you see in John Wick 3 are digital. In the case of throwing knives, we won’t even have a handle — people will mime out the throw, mime the catch, and mime someone getting hit. It’s all mimed out and rehearsed meticulously. They practice this stuff so the timing feels correct. What they’ll do is build up the set with cardboard boxes, and shoot rehearsals. It’s not something the cast usually does, just the stunt guys, but in this case Keanu did that too. He’s an unbelievably insane dude who went to the absolute max. The way we did it was like ballet.

I’m really fussy about data capture. In fact, I’m a pain in everybody’s ass about it. We scan all the performers and all the knives, and get complete, super-photogenic versions of every single person and prop, so we can re-create them digitally. If you want it to look real, you need real-world measurements. You need lighting measurements. You need to do the work.

Game of Thrones, “Book of the Stranger†(2016), Khaleesi burns down the temple

Sam Conway, effects supervisor: Burning the Dothraki temple was a tricky little number. The hard part was all the preplanning. We broke it into an interior set and an exterior set, two separate pieces, and it ended up being shot in two separate locations. The exterior was done in the south of Spain, Almería, and the interior was done in Belfast, in a parking lot behind the production office in the cold winter.

When we did the exterior, we approached a few local gas companies to persuade them to use liquid propane. And then we did a series of tests to please everybody. Eventually the set was constructed out of a steel framework, which allowed it to have some integrity until it burned. And then it had wooden panels and wooden beams. It had a straw roof. We knew these sort of things would burn really well once the effects were turned on. Once all the pilots were lit and the cameras were all rolling, we brought it on bit by bit, section by section, until eventually we got to the height and intensity that was required. It must have gone up about 80 feet. Basically, because of all the straw and all the thatch that was on top, it would carry on burning up with embers in the air. It was a night shot, so it was spectacular. I think you could probably have seen it from quite a few miles away.

One of the camera angles was a shot zooming in onto what would be Daenerys standing at the step. What we had to do there, for that particular moment, was make a safe way for Daenerys to push over the bowl for the fire. It wasn’t gonna hurt her. It was gonna be a very controlled manner, so it landed and sprayed out in front of the camera. We basically cheated the perspective. We had her at a safe distance so she wasn’t actually touching the bowl. She was a good foot and a half away. And as she went to push, we pulled the bowl over with a wire at the same time so it had the appearance of her touching the bowl, pushing it over. We put the flame-bars around the floor, so when the bowl comes down, it ignites them. We must have had about 150 flame-bars in the roof, along the walls. In a production meeting in the early stages, I was trying to explain that shooting in the interior set would be like being inside of a pizza oven. You couldn’t have high flames, ‘cause everywhere you’d feel it.

Die Hard (1988), jump from a roof

John McTiernan, director: Since we were working around a lot of people on a real building, we had to figure out a way to set a fire without damaging anything — and then put it out. So for the explosions, we built these crazy machines that had a great deal of propane in them, vaporized. On “action,†they would make a great deal of fire for about a five-second burst. When Bruce [Willis] jumps off a roof, I think that was a piece of scenery actually built on the roof, but built from the edge, so he could jump down and not jump all the way off a building — obviously. But to put the fire behind him in the traditional way would have been an enormous amount of nonsense. Instead, we had these mortars drive around with a forklift, and they would put fire exactly where you wanted it.

At the time, computer techniques for filming in the background behind something were comparatively primitive, and we almost never had a moving camera. If a camera was moving or was involved in a scene, the audience felt like, “Hey, wait a minute — this is a special effect.†If I remember correctly, the camera goes past him and he looks down. That makes it clear to the audience, “Hey, this is real. We’re all up here on the top of this building.†Then when we started playing tricks later, the audience didn’t notice it as much. Then you cut to some things on sets!

There was another scene, where Bruce swings out and shoots the window, that was filmed on a stage — not actually on a building at all. There’s a background behind him with the lights of the city, and it was a high-ceiling stage, so he could swing back and forth pretty wide. And that’s where he swings out and shoots the window. I don’t think it was traditional green screen, but it was faked in a certain way — just outside the window, there were four or five stuntmen pulling on Bruce to drag him out of the window again. You actually have to have a man doing it, not a machine, so that you can know if there’s a problem and it stops instantly. I think he probably would’ve fallen about 15 feet had he actually gone out the window again.

Adrift (2018), pitchpole shot

Sara Bennett, VFX effects supervisor: Adrift was based on a true story about a woman lost at sea, and our part of the storytelling was to help bring to life the storm sequence and how horrific it was. The hardest shot is what we called in-house the “pitchpole shot.†It was incredibly difficult, I’d never done anything like it. You start off on a wide establisher, and you’ve just got the boat on the gimbal, which they shot, and then we create all the environment around it. Then we come down, we’re on the boat, following behind them, and the drama builds, and in the distance you see this 100-foot wave coming towards them. You can see what’s gonna happen and it’s all very terrifying, and then we travel up this 100-foot wave that starts to break as you get to the top of it, then the camera pans around, and then you’re at the end of the boat, and he’s telling her to get into the cabin for safety. And then we drop with her into the cabin. So we’re panning around with that, with water effects and props floating everywhere, and then suddenly you’re ejected out into this underwater environment, which suddenly becomes very serene and peaceful after the carnage. We see Richard get hit by the mast and then he’s just floating within the water and we gradually see him disappear off into the depths of the sea. So it was kind of working on four different scenes setups in a way, but making it seem like one scene was shot.

The first part of the shot is live-action with the actors shot against green screen on the boat. We created the ocean around that. We used parts of our CG boat to get from part one into part two, when we see Tami fall into the cabin. For the interior part of the shot we worked with a plate shot of a padded green box based on the real interior cabin, which they rolled on set, using a stunt double. We created a CG interior and blended between the stunt double and used a digi-double take over where needed. We also added in CG water and falling props — things like CG books, bottles, fruit, and cans of food, along with churning CG water and splashes, which we added to really help sell the chaos. The final part of the shot where we are underwater, everything was entirely CG including a CG Richard up until he gets hit by the mast, and from there we use the real actor.

The work we did was all really challenging because water is inherently difficult to make look real and not like a visual effect. [Director Baltasar Kormákur] had lots of footage based on storms,  which we studied a lot. The biggest selling point of our storm was the foam and the water breaks on some of the waves. That believability in there is really difficult, so we really studied how waves sort of get aeration underneath the color of the water, how it breaks on the surface and you get this white water, which helps sell this believability. We built a procedural ocean using Houdini Engine that got passed to our animators for initial layout. They could then take the live-action boat and hand-animate the timing, speed, and scale of the waves at each section. For the opening part of the shot, the animation team ended up creating 21 bespoke waves before we hit the big hero wave at the climax of the storm. I think maybe eight effects people worked purely on that pitchpole shot throughout a six-month period.

Scary Stories to Tell in the Dark (2019), the Jangly Man

Luke Groves, global head of production, MR X: We worked on Scary Stories to Tell in the Dark alongside its producer, Guillermo del Toro, who we’ve worked with a number of times. The challenge here was to bring to life complex creatures like the Jangly Man — all on schedules and budgets that were not so big for a movie of this caliber.

Working with Guillermo on The Shape of Water, we learned to exaggerate, augent, and amplify digitally what was being achieved on set in camera. We had an actor, Doug Jones, wearing a suit, and our job was to make sure the audience had no idea what was Doug and what was visual effects. It’s like the Pepsi challenge. I can show you two shots from that movie, and I guarantee you could not guess what was digital and what was practical.

The industry is constantly shortening schedules and lowering budgets. In the case of Scary Stories, we were really up against a release date. A lot of projects these days want to get from camera to screen in six months, and there are visual effects shots that take more than six months to do. So our goal is to make it work on that timeline. The trick here was to set out the scope of the work ahead of time and engage directly with the production, including supervising the work on set, making sure everything we needed was being captured in camera. It’s the only way to get it made.

The Jangly Man was only possible as a hybrid. There are plenty of fully CGI creatures these days, but the hybrid approach gets you the best bang for your buck. The prosthetic team works with the visual effects team, and we embellish a design and take it to life. Merging the practical with the VFX gives you a much higher level of quality and believability. It’s crucial, because
audiences these days are smart. They can tell. People watch Netflix in 4K; we have to get them to suspend disbelief.

True Detective, season one, episode five (2014), guy getting shot

Cary Joji Fukunaga, director: In this scene, we had multiple bloody effects to pull off. The first thing that happens is Hart (Woody Harrelson) comes up and shoots a guy they capture who is on his knees. We set the actor up on the ground, and he is wearing a face makeup piece that was pretty impressive. First we shot him clean. Hart comes up, pulls the trigger, it’s an empty revolver, and he falls to the ground. Then, we set up another shot where we had someone fire an actual blank round that’s three different kinds of exit material. One was just a black bag full of blood, one was a bag full of brain matter, and one was a bag of like hair and bone. It was pretty nasty. Then, to finish off the shot, Felicity Bowring, our makeup artist, had spent the entire night coming up with this prosthetic device that went on the actor’s head that was kind of like if a sunflower had opened up inside of his head. It was just disgusting. Then we filmed him on the ground with that, so when VFX put it all together, instead of cutting away to one of the actors when he fires, we show Woody Harrelson, we pull with the camera right up until he’s raising the gun, and then we turn around and show what he’s doing. You see his gun go off and this guy’s head open up and he falls to the ground. So that was pretty brutal.

And then, his partner turns to make a run for it, and this is where he sets off one of their own booby traps — a “bouncing betty†mine, which is an anti-personnel mine that goes off about two to four feet off the ground. It’s designed to be maiming and killing people, and it goes off at the midsection of the body. The way we shot that is this character goes and makes a run for it, he at some point sets off this mime and will get blown up. And then Cohle (Matthew McConaughey) was aiming to take a shot, and when this mine goes off, he dives for cover, and you see a bunch of pellets from the mine hit the car around him. It just adds to the severity of the effect because it’s saying, not only is this guy blown up by the mine, but even people that are 20 feet away are in danger of the shrapnel.

So the way we shot it was: for the guy running away, getting blown up, we actually shot with special effects a sort of bouncing betty device that would be shot on the ground. Then we did it again, where we did an explosive device at the apex of where that bomb sprung up. Then, we did a blood and guts explosion, and that included a dummy upper body. So then you marry all those things together, plus the red – which is a red mist ultimately – and he eventually disappears in the explosive part of it, and when the flash clears, you see this upper body tumble to the ground. Then on Matthew McConaughey’s coverage of the same moment, we had set up this junked out car that was behind him with pre-rigged bullet holes. Then when we put it all together, we were able to digitally erase the holes in the car and make those appear within a number of frames on the edit. We did the same thing with the car windows that explode behind him. It really gives you a special sound design added in, this kind of peppering of ultra fast metal flying in his direction and tearing up the car, but miraculously not hitting him. So within a matter of seconds you get a pretty brutal series of special effects and visual effects, which should give you a sense of how fast and brutal violence can be and how quickly it is done.

Harry Potter and the Prisoner of Azkaban (2004), time-turner

Roger Guyett, effects supervisor: The first one I thought of is from Prisoner of Azkaban. We did an effect for the Time-Turner, a pendant that Hermione wears so she can go back in time and take extra classes, which becomes very important to the story later on. Working with [director] Alfonso [Cuarón] on that film, he had such tremendous imagination. When we started talking, he really wanted to capture the spirit of seeing time reversing itself, circling around you. It’s done in a pretty elegant way. The shot, designed in part by our cinematographer Michael Seresin, required a tremendous amount of planning, because the sun was moving at a different rate in the foreground and the background. We had the original camera moving with the main characters, and to get everyone speeding up in the background, we had to shoot multiples passes with a motion-control rig. In all the activity happening behind them, we can get everything moving as fast as we like, even though both sets of actors are moving in normal time. It sounds relatively trivial until you actually figure out how to do it. So we ended up using plenty of miniatures — miniature clock tower, miniature grounds — really relying on graphics and planning. It’s a multi-element shot stitched together with digital technology, but mostly, it depends on old ideas.

Midsommar (2019), smashed skulls

Photo: Gabor Kotschy

Vico Sharabani, founder and executive creative director, The-Artery: For Midsommar, it was a huge challenge to design and craft a CG treatment that would depict the physical devastation suffered by a pair of sacrificial offerings. We were brought on with a little less than three weeks before the theatrical release deadline. We created 30 complicated CG shots — skulls being smashed by sledgehammers, faces and limbs and torsos shattering against rocks after being thrown from cliffs above, blood treatments. All amazingly complex treatments within an incredibly short window of time.

This type of creative development generally requires a lengthy process of research and development, as well as a significantly longer period for execution. But with the release date looming, production reached out to ask for our assistance. In the end, we were able to deliver with a great number of hands on deck and working 24/7 to get it turned around in time.

Jurassic World: Fallen Kingdom (2018), gyro-sphere escape

David Vickery, effects artist: A lot of my job isn’t figuring how to do visual effects, it’s figuring out how not to do visual effects. We’re trying to find ways for our directors to capture real action in real spaces, without relying on fixing things in post-production. For the gyrosphere escape, there was so much VFX content in that scene alone: the volcano explosion, all the dinosaurs stampeding as they flee, and the vehicle rolling down the mountain.

So we built large gyrosphere rigs that would allow us to put Bryce [Dallas Howard] and the other actors into the plastic bubble that we could pilot remotely across the undulating terrain. And to capture that, we had off-road pursuit vehicles with cameras mounted onto crane arms. In post-production, we’ve got the not-small challenge of painting out the rig parts and replacing the glass around the outside of it, but the actors are bumping around just right — if they go over a pothole, they get a nice big kick thanks to the suspension.

Another task was finding a way of controlling the weather, essentially. Chris Corbould, our special-effects supervisor, worked out where we’d need to lay hundreds and hundreds of feet of fog-and-smoke tubing. We brought in huge industrial-grade foggers and essentially diffused the light and atmosphere within the outdoor space to simulate consistent weather. So the gyrosphere is bouncing along the ground, and we’ve laid hot charges to detonate that give us explosions — like lava bombs — raining from the sky, which helps the editor once he’s cutting, because now he’s got reaction shots of actors freaking out as they’re being peppered with bits of mud.

There’s a bit where the gyrosphere goes off a cliff, and we thought, “Can we really put Bryce Dallas Howard in a plastic ball and push her off a cliff? Probably not.†But in one beautiful moment, our director J.A. Bayona said, “Well, why don’t we shoot it at a theme park?†Instead, we built a 50-foot-tall, magnetic, custom roller coaster in the back lot at Pinewood Studios. A team of engineers used that coaster to work out exactly how that moment would play, and we just strapped our cameras that rig. We spent many hours rehearsing with stunt doubles, but saved the first take Bryce and Justice Smith did, to get the genuine moment of terror.

Solo: A Star Wars Story (2018), Kessel Run

Rob Bredow, visual effects supervisor: We developed the whole climactic Kessel Run sequence as a single continuous experience, both for Ron Howard to direct and for the actors to live through in the cockpit. We wrapped an entire 180-degree screen around the cockpit of the Millennium Falcon, and had multiple projectors throwing media onto that using rear projection, one of the oldest tricks in the book. But we used new technology from ILM, creating an entire world of the Kessel Run in high resolution to be put on those screens. When the actor pulls the level to go into hyperspace, they really see the stars bleeding out like that.

What I didn’t realize that Ron would want to do, is stage the entire Kessel Run end-to-end live. So it was a 20-plus-minute scene, with something like 76 different lighting and media cues we’re all working out, plus we’re standing right next to the director of photography Bradford Young and he’s working out his own cues with his gaffing team.

This was shot at Pinewood Studios, with the Falcon itself about 12 or 15 feet in the air on a gimbal we could use to bounce and spin it around. Ron liked the idea of doing long, continuous takes so the cast could get into character, and so he could get lots of footage to give himself options. We realized this Star Wars film would have more taking place in the cramped space of a cockpit than we’d ever shot before. There are only so many ways to shoot such a compact space. One of the fun things that rear projection allowed was a fresh compositional element: The shot where Han first walks into the cockpit, with L3 and Lando sitting at the controls, you see Han’s face light up when they jump into hyperspace. Better yet, you can see the reflection of the stars in his eyeballs. We got that shot in the moment in the cockpit — that’s just the rear projection being reflected back.

As a collaboration between VFX and camera department, it was really fun too. We’d be sitting there, and Bradford would say, “Oh, that blaster fire, can that be backlit?†We could make these changes on the fly that would normally happen in post. The blaster fire, which we nicknamed “pews†— because, you know, “pew pew†— those could be moved, changed in frequency, color, shape. Writer Jon Kasdan would send us new pages at two or three in the morning — we wanted to be ready for whatever came our way. It was pretty fun! Felt more like putting on a live show than doing traditional visual effects.

CSI, “Post Mortem†(2006), serial killer’s dollhouse

Andrew Orloff, visual effects supervisor: A lot of the work that we did on CSI, back when it was the hottest show on TV, was really interesting, because it was a mix between the newer CG techniques, and all computer-generated techniques, and the older generation of computer-controlled cameras and motion-controlled cameras. So we would do a lot of shots of, like, diving into people’s bodies. We’d have a mannequin of a human body, and we’d take a special snorkel of a camera and program a robotic arm of a camera to do the same passes over and over again — one pass for the body, one pass for the blood, one pass for the transition.

Once, we were doing a shot where there was a serial killer who re-created all of his murders inside of a dollhouse. We wanted to have the real-live action take place inside of the dollhouse, with the same camera moves, and the same people appearing. So what that meant was photographing a bunch of material in live-action, and then photographing the dollhouse scaled down on a different motion-controlled camera rig, so that technically, it was a good amalgamation. I love when visual effects are really involved in the storytelling and really involved in the more technical parts of filmmaking and camera work.

Even though it may not be the flashiest effect, or the most Emmy-winning effect we’ve ever done, I learned the most. When you’re grounded, when you’re taking input from all of these other pieces of the filmmaking process, and you’re using visual effects to integrate — to play jazz off of all of the different artists and people who are making this image on film — it becomes something that’s more than just a visual effect. It becomes part of the story.

The Ballad of Buster Scruggs (2018), carrying man up stairs

Alex Lemke, visual effects supervisor: For me, the hardest was Liam Neeson carrying Harry Melling up the stairs in The Ballad of Buster Scruggs. We had to find a way to do that because he’s not supposed to have any arms or legs. So, we had to come up with a super-complex, motion-controlled, motion-based setup to make that work.

Michael Huber, visual effects supervisor: It looks just like somebody carrying someone in a backpack, but in the end, it’s a very complex shot. We knew from the beginning that there was no way to have Liam Neeson carrying a real person on his backpack, so it took us a long time to come up with an idea that was practical, but at the same time, a guaranteed result.

Lemke: There’s also the second part of the shot where he puts him on the floor and turns him around — and even though you try to prep everything, and we made storyboards, this was kind of improvised on set. I think at some point, there was a talk of, “Can we actually do this in one take or do we have to have a cutaway?â€

Huber: In the wide shot where he carries him, we might have gotten away with a digital hand because it’s further away, but then when he puts him on the floor, it’s a real close-up. Digital humans are a tricky thing and it can turn out really weird. So, we decided that the head is always real actor, but the body would be a mixture of a dummy that’s a backpack with a body on it  — and then somewhere between the shoulders and the chest, we are actually dissolving into the real Harry.

Lemke: We shot the real Harry onstage, then we did a 3-D model of the backpack, and used that model to generate the motion that was present in the original play. That looked really good, but what that meant was, we would have to juggle a lot of systems, and there was no way to shoot it in an easy fashion. So we put the translation on top of everything else, so Harry’s prosthetic looks like it’s moving up the stairs. The camera moves backward to make it seem like he’s moving up the stairs.

Huber: The rotation is Harry really rotating, but the translation is not him — it’s the camera doing it instead. All of that has to be programmed and it has to be calculated. On top of everything, we had to shoot it half-speed, and then speed it up in post. It’s complicated, but it just looks like a guy being carried up the stairs, and that’s what it should be.

Stranger Things, season three (2019), The Goop

Martin Pelletier, VFX supervisor, Rodeo FX: The Duffer brothers are full-on fans of ’80s movies with old practical SFX — good old animatronics. If you want fake blood, you bring fake blood to the set, and if you want slime, you get real slime. Of course, this approach requires really long reset times, and it’s not always convenient. The main reason they prefer it to CGI, though, is if you look at CGI creatures versus those in, say, The Thing or the original Hellraiser, the CGI monsters aren’t dripping goo; it wasn’t really possible to do that with CGI. So going into Stranger Things three, the number one thing they said was, if we are going to make a creature in CGI, it has to be insanely wet and slimy and stick. Everything has to drip.

The first creature we did was the first version of the Mind Flayer — this little creepy thing that forms after the rat explodes inside the cage, in episode two. We refer to this creature internally as The Goop. The Goop would be the first creature established as the materialization of the Mind Flayer, and it would grow and evolve and become more agile and defined. We started with the idea of thousands of rats screaming in pain in a very dark basement. The Duffers wanted the rats to explode — or to be more precise, implode. That made it a little more interesting. You’re not looking for guts and blood to splatter all over the place like a grenade effect. You want the rats to sort of flip inside out, instantly. It would splatter, but not separate into isolated chunks.

The problem was, as much as the Duffers loved gore, they said they would have a hard time getting it through Netflix if it was too descriptive — that is, once the rats start flipping inside out, we had to be careful not to make it too biological, where we can see the heart and the lungs and the eyeballs. It had to be something that looked nasty, but on that thin line between spooky-cool and outright gory. So we made a system using an animated rig rat asset and a recipe from the creature effect department. We could make it flip inside out and hide the fur instantly in a thick pile of nondescript guts and blood and mucus.

Doctor Strange (2016), Strange’s psychedelic journey

Stephane Ceretti, visual effects supervisor: Doctor Strange was complicated because the world was magical, and everybody had an opinion on how it should look, and you try to accommodate everything, you try to find the best way to make it work, and because it’s not based on reality, it’s really hard to actually land on something that everybody’s happy with. It’s so different from what they were doing before, so it’s a lot of discussions. But they’re all worth it. One of the most challenging sequences was the Magical Mystery Tour, which is the moment at the beginning of the film when the Ancient One is pushing Strange out of his body, and he flies across all these dimensions. It was very difficult from a conceptual point of view, because it was an open thing. We knew what we wanted to tell, in terms of what was written on the page, that the Ancient One was explaining to Strange that there’s so much more to the world than he knows, but apart from that … People had opinions about how long it should be, what we should show, you go through every single beat and you go through a lot of image references and ideas and suggestions from everyone, because everyone’s going to chime in, and then at the end you just need to get it done.

It wasn’t very detailed initially on the page, it was mostly kind of a voice-over scene with, you could hear the Ancient One talking to him, and so that was what was on the page, and there was some very ambiguous descriptions of the world he would be visiting. We knew that we wanted it to go through the quantum realm and introduce a dark dimension that we’d see later in the film, and these were the main beats, but in terms of what it is and how it looked and what was really happening to him, it wasn’t so much in the script. But that’s when we start just diving into it and getting ideas up and looking at inspiration a little bit everywhere, in photography, in painting, in sculpture, some scientific experiments, some microscopic photography, quantum whatever. We just kept looking everywhere. It was a very complex shoot, because we had to put Benedict in what we call a motion base to film it, and we had to invent the motion base to do what we wanted it to do, to make him spin around and fly in all these dimensions. It was very difficult on a conceptual level, and it was very difficult on a technical level, because we had to have the stunt and special-effects guys involved, and we had to have the DP involved in the lighting of it all and we had to have motion-control cameras, and then finishing it and making a sequence that made sense for the film was very difficult as well. Every step of the way, there was a challenge. We started so early in the process, and we finished it last minute, so it took us two years, pretty much, to do.

Starry Eyes (2014), dumbbell scene

Dennis Widmyer, director: I think “hard†is the wrong word, but it was brutal. We wanted to build a mold of the actress’s head, and we wanted it to actually break apart realistically, so SOTA FX actually put bone underneath the mold, so there would be something to break. Alex Essoe is using a real dumbbell, so she had that dumbbell and another dumbbell that had a blood pump on it.

Kevin Kolsch, director: What really made the effect hard wasn’t the actual play of the effect, but the aftermath. We only had one head mold, it was going to be destroyed, and there was going to be a lot of blood going, so there wasn’t going to be any second takes — and that’s one thing that’s kind of nerve-wracking, doing an effect, is when you know you only have one shot at it.

Widmyer: The number one reason that things don’t always work with special effects is just that the blood doesn’t flow the way it should. It’s either too much or too little, or just doesn’t work at all. So they used a fire extinguisher filled with blood, which is the ultimate way to get blood to spray and pump. But we might have gone a little too crazy with it, because when we actually did the gag, it worked so well that it sprayed the entire room. There was blood all over the ceiling, all over the drapes, all over the walls, the blood seeped down so deeply into the mattress that the actual guy whose bedroom it was — because we were using a buddy’s house — a week later, we still couldn’t clean it out.

To this day, he never got the blood out of the box spring.

Kolsch: We were using our buddy’s house, not a set, and being a low-budget indie film, this was not a movie where you shoot and then you go, “All right! That’s a wrap!†and you turn around and walk away and in comes your locations department, and your set dec people, and they start cleaning up. No. It was me, Dennis, and Travis Stevens our producer on our knees until four in the morning, like in the kitchen, mopping up this blood.

The Hallow (2015), basement creatures

Corin Hardy, director: When I was making The Hallow, there were loads of challenging effects throughout because it was a real ambitious attempt at doing everything as much for real and on a low budget. We tried to make it as practical and in-camera as possible, and use visual effects sparingly. So we did full creature suits with animatronic heads, puppeteer-able arms, like a five-stage prosthetic transformation body horror, full-size infected stag, an animatronic stunt baby — all using a seamless combination of practical effects and CGI, stop-motion microscopic infections. Some of it worked out better than others.

Toward the end of the climax of the movie, Adam — who is infected by one of the creatures and is changing into one of them — goes down into the basement of a house and discovers where the creatures dwell. It was really a complex scene in full body-covering prosthetic on the costumes, and people crawling around in the dark with small breathing tubes to allow them to breathe in between takes. Two of them had full animatronic heads on, meaning that their heads are essentially shut inside of a heavy mechanical one and each is operated by puppeteers. Two of them had puppeteer arms that they could extend themselves as they crawled around, and each of the performers had a special skill, so we had like a contortionist, a ballet dancer, a mountain climber, an animal movement specialist, and a gymnast — it sounds like a joke.

So, they’re lurking around the basement, which was in a studio, and it had become really hot on account of Joseph Mawle, the lead actor who plays Adam, whose arm has real flame powered by a hidden fuel tank. He’s also sporting an eye and shoulder prosthetic, so he can’t really see very well or hear very well. Then we also had Iggy, his dog, which is also changing into a Hallow, who’s been wetted down and covered in prop slime and instructed to lay still whilst this carnage happens, and most difficult of all, we had Cora (Charlotte Williams) who is an actor in a full body prosthetic with a fully animatronic head that’s being controlled by three puppeteers off camera, and in her arms, she’s carrying a full animatronic baby who is also being operated by three more puppeteers. So, put all that together, and the thing that made the encounter and the effects so much harder was that this was intended to be an incredibly emotional beat in the movie. I wanted to get this subtle dark tenderness in Mawle’s performance amid this dramatic insanity, but all while fire and prosthetics and breathing tubes and animatronics are around — and trying to do it all safely. I think we shot it all in one day, which is crazy. It taught me not to be so silly as to try to ever attempt to film dogs, babies, and full prosthetics of an actor in flames in one day. No, I’m kidding. It was fucking hard, though.

Isle of Dogs (2018), water

Tim Ledbury, effects artist: Isle of Dogs was unique in terms of how we wanted to approach it, but I think specifically the hardest thing to work out on the film was how to do the water. It took a lot of VFX work, as well as practical mechanics, so there were many systems we used to achieve that effect.

Primarily the challenge was, we were not allowed to use CG. Well, I mean we were allowed, but we had to create this effect for real. We tested hundreds of different times in terms of how to do the sea, but in the end we ended up with this sort of mechanical system with many layers of rubber and hair gel and saran wrap on top of this mechanical base. Then, we had to photograph it with a camera very close-up, because we had to composite into the shots since it was very miniature. So, the challenge was trying to shoot all of these elements that would then go together with a relatively bigger puppet, and combine these things together so it would look like it was shot in the same world.

If you put it all together, it was probably five months of shooting water elements, which is an extremely laborious process. At one point we were shooting four whirlpools and we had four animators all focusing on their whirlpool. It takes a long time to do stop-motion. You don’t do it for quick results.

Mortal Kombat (1995), Goro

Tom Woodruff Jr., director and effects artist: The hardest effect I’ve ever done always goes back to Mortal Kombat with Goro, the four-armed creature. Because, let’s see, when was that? That was 1995, and it was all so new, and it was a [tall] order to fill because the producer knew that he wanted it to be animatronic — this was before digital was at a point of working very well — so, the director wanted the flexibility of having an actor — me — inside of the suit. We had four different puppeteers on Goro’s face because he had to lip-synch, and we worked with a new system of prerecording his lip-synch so that he was always on-synch on set when he was interchanging dialogue with the other actors.

There were two versions of Goro. There was the stunt version and what we called the hero version, and here’s the guts of the hero version: my head was inside a black dome and the upper arms were operated with a high-speed motor that we had to develop in order to make them move. And in order to get those to move, a puppeteer had to put on what we called our “telemetry suit,†where his joints matched up with the joints on the upper mechanical arms, so whatever he would do, the upper arms would do the same thing. The downside was that the whole package that I had to wear was 120 pounds, which wasn’t bad when it was sitting up on my shoulders, because I could kind of displace some of it down to my hips, but the minute I leaned the slightest bit off-center, which I had to do for a lot of the fighting, the weight was always against me, so no matter which way I turned, most of the performance was essentially weight lifting. And then, because there was no place to hide eye-holes on the chest of Goro, we had a system where I had a little video monitor inside and I could see the image of what the camera would see, so the camera would transmit a picture. So I would be able to see what you’re going to see on-screen, and I would know where to stand and make sure I’m in the light, and how [to] turn so I’m facing the actor the right way and reacting to things. So there were a lot of moving parts that had not been tried out.

We were also set with a very, very strict, tight budget. There are times when things go slow, there are times when you have a little breakdown of something, swap out a motor, but in the end, people have loved this creature. And it happened so many years ago, I’m over the pain and the problems of it. But it was interesting that a little while ago, Richard Taylor, who is Peter Jackson’s lead supervisor, said that when he and Peter were young, they fell in love with Goro and they couldn’t figure out how it was done. He said they would buy videotapes and wear them out, just fast-forwarding to the Goro stuff, and I thought, “Wow! That’s cool!†And I keep waiting for that phone to ring, for Peter Jackson to call up and say, “You guys build something for me!â€

The Umbrella Academy, “We Only See Each Other at Weddings and Funerals†(2019), the dollhouse dance

Mahsa Ghorbankarimi, CG supervisor, Spin VFX: One of the most memorable scenes in Umbrella Factory is when we see the main characters dancing alone in their rooms and the camera zooms out to reveal them all in the same house. It was very interesting on a conceptual level, because it had to have a dollhouse look, and look like a recognizable dollhouse, but at the same time needed to look completely realistic, like a house that has been sliced in half. It needed to feel like it was set in real space so it could become a visual reference for the show, knowing where the characters are living.

The camera move was extremely hard to achieve. We had to start with a plate of Ellen Page that was shot on set. She’s dancing in the foyer of the house. Then the camera pulls back very slowly, to the point where it reveals the whole house and all the rooms in it. We had to combine several different shots, but we didn’t want any strange camera distortions.

The challenge lies in the fact that we have eight different characters dancing in eight different rooms that were each shot on set. Each was shot with different cameras, from different perspectives, on different lenses. A lot of compositing work went into just making them look like they were the same shot. We needed to place each image in its respective quadrant in the house, then run through a CG camera move, finessing the individual shots so they work with the perspective and the new parallax movement the CG camera was creating. We had to build a CG version of each room, and then project the shots onto the CG assets, and then matte paint the parts of the shots that were breaking down due to the parallax movement of the camera. And then we had to light all of the CG rooms to match the real light and cast the correct shadows in the correct places.

Finally, we had to separate the dancing characters from the shots and composite them back in the house after it was fully rendered, making sure each person was in the right room. The end result is a short shot — it lasts maybe five seconds — but it created an iconic look for this space and it’s memorable throughout the entire series.

Herbie: Fully Loaded (2005), 360 shot of Herbie

Angela Robinson, director: The whole finale was Herbie at a NASCAR race at California Speedway. There was only one race at California Speedway, so we had to get everything on the one day, during the live event. Which involved 28 cameras shooting, and it involved putting Herbie on the actual NASCAR track. And it involved bringing Lindsay Lohan to be in front of the crowd. The finale shot was Lindsay standing out of the sunroof, on Herbie, with her hands in the air, winning. And we do a sweeping 360 around Herbie and Lindsay, in the middle of the ground, in the middle of California Speedway, to be the big, glorious, spinning, “Yay, we won!†shot of the movie.

In order to get the swooping shot, we couldn’t bring the crane into the middle of California Speedway and do the many takes that it would have taken. We basically planned it like a military operation, where we had about half an hour before the real race began to kind of get all our footage in. So we had 28 cameras, and, what we needed to do was to get a plate shot of the background of the California Speedway, and then recreate Lindsay in it, and Herbie in it, on a soundstage, months later, in Burbank. So we shot the background of the shot first. All the cars, all the things, all the fans, the, you know, hundreds of thousands of people. And then, I as the director was like, “Good, that’s in the can,†promptly forgot about it. [Laughs.] ‘Til months, months, months, later, when we had to shoot the Lindsay and Herbie part of it. That involved being on a soundstage at Disney, in Burbank, with a 360[-degree] cyc, just green everywhere. And there was this huge robot arm, that had been programmed to mimic with the camera the whole length of the continuous shot that matched the background plate we shot at California Speedway.

It was something like a minute and 12 seconds, but there wasn’t any roaring crowd. So basically, Lindsay had to stand on a totally silent soundstage, in the middle of Herbie, with her arms lifted, beaming with the joy of winning the race in the movie — while this robot arm went “bzzzzz†really slowly around, in the longest minute and 12 seconds that I ever lived. [Laughs.] And then to add to that, Lindsay got some very upsetting news moments before we were to do the take, and was upset. And I went over … and this was like … so much money had been spent just on this shot, because we were renting the robot. So much preparation has gone into this one thing. And there was nothing to do. So she was such a rock star, and such a trooper. I can’t say how amazing she was, because in her heart she was really upset but I was like, “You have to manifest pure joy and happiness for a minute and 12 seconds, ‘til we get this machine to go right.â€

So, she got herself in there, and I went, “Action!†And she manifested pure joy, while this robot arm went “bzzzzz†around. And then I yelled, “Cut!†And then, she was still upset, so the emotions overtake her. I sent her back to hair and makeup. She came back, I was like, “Let’s do it again.†[Laughs.] So I think we did 12 takes, and it was one of the most surreal things I’ve ever done, where it was just standing, the antithesis of the thing.

Welcome to Marwen (2018), dolls

Kevin Baillie, effects supervisor: By far the hardest part was creating the doll characters themselves and delivering on Bob Zemeckis’s edict that the souls of the actors have to shine through in doll performances. Right off the bat we said, we don’t want to use traditional motion capture like we did on A Christmas Carol. We need to invent something new. And initially we planned on filming the actors in costume just like a live-action movie, and then using digital tools to shape their bodies into the more heroic doll-shapes, and putting computer-generated doll joints on them. We shot a test of Steve Carell using that method and it looked terrifying. It was horrible. It just looked like somebody in a really fancy Halloween costume. So then we tried to doing a motion-capture version of it. For all of the reasons we thought motion-capture wasn’t a good solution, it proved not to be a good solution.

Then we got this idea to say, well, why don’t we try to create a digital Hogey [Carrell’s character] with Steve parts? So we used pieces of his eyes and his mouth from actual filmed footage on a digital doll. So the end result of how we ended up tackling the dolls was that the dolls are totally digital except for their eyes and mouths, which are actual living, breathing footage of the actual actors. And figuring that out that was hard and then executing it for what ended up being 46 minutes in the movie, which you spend in the doll world, that was also challenging. ‘Cause doing it as a one-off is easy. Doing it for 46 minutes of a movie is a whole different ball game.

C. Kim Miles, cinematographer: Instead of just a traditional motion capture in which actors in gray suits run around and act out the scenes, we introduced our traditional filmmaking equipment — dollies, cranes — into that motion-capture environment, which by my understanding was one of the first times anyone had ever done this. So for example, if we were in Captain Hogey’s little bar, we lit the the motion-capture stage so that a lamp on the bar would interact with an actor’s eyes the way that it would in real life. The trick with that was that our movie lighting tends to emit a lot of infrared energy and the motion-capture cameras that are looking for the movement information from the cast members are infrared sensitive, so if we put too much light into the set, the motion-capture cameras would lose information because they’re being overwhelmed with the amount of light. So we had to scale our traditional lighting methodology down in a way that still allowed us to achieve the contrast ratios that we wanted for the scene, but allowed the motion-capture cameras to still work. The end result is real faces on digital bodies, but the digital bodies are driven by actual human emotion.

The Vampire Diaries, “Bad Moon Rising†(2010), werewolf

Julie Plec, executive producer: Okay, so basically this is the story of how we began our werewolf adventure on The Vampire Diaries. [Laughs.] So I was not there when we shot this episode — the first time. We had met with some live wolves, which for safety reasons are actually wolf-shepherd hybrids. And so we had picked out the live animals, and then I went back to Los Angeles and minded my own business. And then when we got the cut, it was such a joke. [Laughs.] Because basically we were just relying on a bunch of shots of the wolf in the woods, trying to make it look scary, when all it really looked like was a dog. So we went into panic and crisis mode, and did what we should have done in the beginning, and handled the wolf unit like an entire unit of its own. And I called Greg Nicotero, who I had worked with on the werewolf movie Cursed when he was still an effects guy. He and his company had worked with Wes Craven for years. [Plec was once Wes Craven’s assistant.]

We called on him and said, “Please come over and help us out.†And, “What do you have that’s fake werewolf stuff that we could borrow, and help us operate this whole unit?†So he came in and directed for a couple days. We did a huge green screen unit where we put the dog-wolf in front of a green screen, just to basically try to get it to bark, or growl, or look, or do anything. So that we could then drop that into a woods shot. He shot all of our plates for us out in the woods. He brought what we call Stuffy, with a capital S, which is a huge stuffed wolf puppet thing. Basically a stuffed animal. So there are points when literally the shot is the stuffed animal getting thrown into the frame, and then the actor rolling out of the frame with the stuffed animal. He brought a mechanical wolf head that — the scene where Candice [King] is fighting, when it’s on top of her and it’s really close, that’s a mechanical wolf head that Greg brought and let us borrow, dripping with goo and drool and slime. And just shaking it really hard in her face. [Laughs.] And then — and keep in mind this is 2010, right? So for the VFX to build an entire wolf, and to do what we wanted it to do would have been extremely expensive, and extremely difficult. And so that was not an option, or so we thought.

So we were trying to do as much of this, practically, as we could. So we went from trying to shoot something with a dog in a day and it became three days of reshoots, an entire green screen unit, a plate unit in the woods, a stuffed wolf unit, a mechanical head wolf unit, and even then when we cut it all together, we had to basically give the werewolf vamp speed, vampire-speed abilities. Because when the dog turned, or ran, it needed to look cool. So we were like, “All right, we’ll just ramp it up. Make it look as cool as you can.†So we broke all our own rules of what a werewolf really could have technically been capable of. We’re like, “Whatever looks cool, fuck it.†[Laughs.]

Avengers: Infinity War (2018), Thanos

Victoria Alonso, producer: The hardest special effects we’ve done was by far for the character Thanos, played by Josh Brolin. To convey what we had been teasing for a few movies, the biggest villain of all time — and to have that not be an actor’s flesh and blood, but for it to be a CGI character, it involves a lot of work. And for it to not just be a supporting, but a lead character — pretty much, if it didn’t work, then Avengers 3 and 4 wouldn’t work.

A lot of the technology involved creating and capturing Thanos “on screen,†which was helped by the visual effects team using sensors and trackers, allowing Brolin’s performance to come through thanks to the machine-learning facial recognition program we used. This process also allowed for many characters to be on camera at the same time, not just Brolin acting alone as this fictional “creature.†We don’t want to use flashy tools as a gimmick. The way we do every movie is, if there’s a story that calls for it, we will do it. So we’re not going to use the latest and greatest facial capture just because it’s out there. It’s, “Does the character need it?†If yes, then we’ll use it and perfect it. And when the technology the character calls for doesn’t exist yet, we go ahead and create it, like in the case of Thanos. It’s important to understand it as a partnership, not just a stunt. If it doesn’t serve the story, then it doesn’t serve our cinematic universe and is threaded throughout in a way that fans do appreciate.

From Beyond (1986), Pretorius creature

Mark Shostrom, effects artist: Pretorius was an ambitious creature because of a couple things: One was that, mechanically, it had to work. And this was in the days before CGI, so it was basically a guy in a slant board, inside this rubber mechanical creature, who had to operate two mechanical hands. Puppeteers operated the legs and a puppeteer also operated the neck. And then we had radio controls for the mouth and radio and cable controls for the mouth and the eye blinking and the eye moving. And all this for, like, I think it was $40-45,000.

With the Pretorious creature, there was the challenge of making this mechanical creature that had to function on set and mime the dialogue, and go through three different facial transitions. Of course, for the dialogue, Stuart [Gordon]’s idea was to cut in close and have the actor, Ted Sorel, in matching make-ups. Which involved actually gluing Ted in with the rest of the mechanical creature behind him, so essentially this actor is wearing this entire creature behind him. He’s glued to it; he can’t get up and go to the bathroom. The monster sculpture, when it was being sculpted, was probably 2,000 pounds. You can’t pick that up and take it and show it. So much was riding [on] Dave Kindlon’s mechanics. And he did it all himself. I said, “Dave what do you need?†and he said, “Well, we need a mill.†I said, “What’s a mill?†“We need a big drill press.†And I’m like, “One you get at Sears?†He goes, “No, like a big fucking drill press. And I need a lathe.†And I’m like, how much are all these things gonna cost? A mill is only $2,000 and a lathe is $3,000 and I’m like, “Ka-ching, ka-ching.†But I said, “No, I gotta do it.
There’s only one way to do it, and that’s the right way, and it’s expensive.†In the back of my mind I’m thinking, “If this works, this could really help my career,†and it did. So, it was all worth it, it made everybody happy.

Channel Zero: The Dream Door (2018), cat steps out of a door

E.L. Katz, director: This was the most challenging scene out of the six episodes of Channel Zero: The Dream Door, because we didn’t have a lot of time and we didn’t schedule lots of time for the simple act of having a cat step out of a tiny door in a wall. That’s literally all this cat had to do. The characters are standing in front of a door that we installed in front of a bathroom, there’s a cat handler inside, and they’re supposed to give a speech, lean down, open a door, and this cat is supposed to come out of the door. It just has to emerge and be in front of the camera long enough for the camera to register it to some degree. It took us four hours to get anything.

A little backstory: it’s a Lykoi cat, which is a super rare cat and they look sort of like werewolves. The whole point of it is, it’s a play on words — the guy is saying, “Show me a werewolf,†and the other one says, “I’ll do a cat;†he says, “Okay,†and this werewolf cat is supposed to emerge. The thing about these fucking cats is, they’re feral, so there was a chance that they could attack people if they jumped out of the door. Also, it’s a very distinctive wolf-like Universal monsters sort of look, so it would go to shit if the cat got too nervous because it would lose all of its hair, so the gag would be lost. So, when it was being transported by airplane, we were really nervous that it might get stressed out during the flight and go completely bald and then it wouldn’t make sense when it came out because you wouldn’t know what the fuck it was; or when it’s on set it could get freaked out and start shedding its fur. Either way, that didn’t happen, it arrived on time, we put it in, like, a hotel or something, it’s got VIP treatment, we’re all set up, we’re basically waiting for the scene. We have the cat handler back there, door opens, cat doesn’t come out. We’re like, “Okay, fuck, do it againâ€.

Each time, we have the actors delivering all of the lines because we want to do it in a wide shot, so over and over again, trying to keep the energy up, the actor delivers this monologue, does this whole thing, leans down, opens the door, no cat comes out. We’re trying multiple different things, the cat’s owner is on the other side of the door trying to coax it out, putting treats in eye-line of where the cat’s going to be, but it’s just in the darkness. You can’t see it, it’s not moving. We then get one trainer on one side of the room, one trainer on the other, they’re both making weird sounds, they’re both talking. Finally, the cat zips out of the room, super fucking fast, and runs under the bed, and that’s not really a scene, because you can’t see what the hell it is. We start trying it again, the cat runs out of the room, it’s getting lost, we have to find the cat. Meanwhile, this is a TV shoot, so we have no fucking time, and we’re just like, “Oh my god, we’re not gonna do this scene!†I mean, we can’t CGI this cat, it’s gonna look fake as shit, especially because it has this, like, star moment where you’re focusing on it. You don’t want to build up this thing and a VFX werewolf cat comes out, that’s also going to be like, “What the fuck is this thing?†So, basically, at the end of the night, we’re running out of time, we’ve created this series of clicks and comforting sounds and all these things, trying to create this safe environment for this cat to be drawn out, and we wait, and this is going to be like the last take we’re going to do. The guy leans down, opens the door, and it’s just the cat’s ass jutting out of the darkness. It just stays that way for a while, and then it turns around and slowly walks out, and that’s what we went with. Luckily, we were able to later VFX in a way that wasn’t too obvious, we just split-screened it so it didn’t look like the cat went ass first out the door. It had its emergence. But out of everything I’ve ever done, this was one of the most frustrating things, because there was absolutely nothing we could do. There’s nothing you can fucking do. It’s a fucking cat.

Creepshow (1982), Fluffy the crate creature

Tom Savini, effects artist: I’d never done an animatronic creature, and I had to build the crate creature for George Romero’s movie Creepshow. There’s another FX guy named Rob Bottin. Good luck trying to find him. He changes his phone number every week. He’s kind of a mysterious Lothario, but he’s one of the greatest at special makeup effects. He’s the guy who did all the stuff in The Thing and Total Recall and The Howling.

I called him and asked him how to do it, because he did the animatronic gorilla for a movie called Tanya’s Island. He spent two and a half hours on the phone explaining, step-by-step, how to do it. That’s the way I learned how to make an animatronic creature, and that turned out to be Fluffy. It wasn’t really advice. It was a lesson, a step-by-step lesson, on how to make a cast of the actor. I already knew how to do that, but after making the cast of the actor, you have to produce a fiberglass armature, and the armature holds all the mechanisms that make the face move. The face is foam latex, so if I wanted to make him smile, there had to be cables to make him smile, to make him frown, to make his eyebrows furrow.

I put rubbers, prophylactics in his cheeks to make his cheeks expand and contract like he was breathing. I had tubes behind his teeth [that dispensed] glycerin to make him look like he was drooling all the time. That was the main creature, and his teeth would cut you. But there was a puppet creature, as well, which had rubber teeth, so when he would attack Adrienne Barbeau or grab somebody on the face or the neck, the rubber teeth wouldn’t hurt them. The real sharp teeth were in the main creature. He was the one that popped out of the crate and for scary close-ups. Sculpting his face took about a month and a half. His body was mostly built by the costume department. I casted my assistant, whose name was Darryl Ferrucci, to be the monster, because he would be there with me all the time. Instead of waiting for an actor, he and I did all the effects in Creepshow. We cast his body, and the costume was built on a plaster cast of his body. All he could do was move the head and open the jaw. All the other facial expressions were operated by us off-camera with cables that made the mouth smile, the brow furrow, and pumped drool behind the teeth. It was pretty complicated. It took four or five people to make Fluffy look alive.

Deep Impact (1998), tidal wave

Scott Farrar, visual effects supervisor: Deep Impact was the big first motion picture where anybody really tackled computer graphics with huge water effects. We had this giant wave that was coming to hit NYC. And it was so difficult — we couldn’t render things. We can do a lot of shots that run overnight these days, but some of these shots took weeks to put together, and we were never sure if they were even going to look good.

Even something as simple as having a comet trail — if you have a lock-up camera, things are much simpler. You can put a comet head in that comes flying across the screen, or comes toward the camera. But if the camera is moving, like on a camera car, following along with this thing moving through the sky, and panning and tilting in computer graphics terms, is much harder because you’ve got to lock it into a 3-D space and you’ve got dimensional change all the time. And this was the first time that we were doing all these things. I just remember the pain of that — something as simple as trying to make the Statue of Liberty look good with its copper patina, without making it look CG-fake.

We shot this scene where the waters were rising up higher and higher, and the kids are running away in the forest, trying to get away. And finally the water starts to recede. They shot in L.A. — they were on top of the ridge, looking across the valley. And we were trying to put all this water receding, and see the damage that’s left. Now, we could do a better job of that: just the rotoscoping, and the tracking, and the matte paintings — everything could have been better today. You see water effects in everything now. But that was before The Perfect Storm. Remember The Perfect Storm? Where there was a giant wave? I worked on that when it was being developed. So that one benefited greatly from what we had done in Deep Impact.

Agents of S.H.I.E.L.D., “ASCENSION†(2015), Hive’s true face

Mark Kolpack, effects supervisor: It’s always a mad dash to maximize everything you can in order to pull off the super-difficult work that maintains the quality of the Marvel brand. The most challenging thing we’ve done, which turned out exceptionally well, was a full head replacement with facial capture for a character named Hive. It had all these tentacles. We had to get the right textures mapped for his skin tones, and get the suckers on the tentacles right, so they all felt like they had different sizes and feels to them. All those details probably took us about four months, from the initial design of it to the completion of it.

What made it extra challenging is that we had some very complex on-set lighting. And Hive’s head was taller than Brett [Dalton], so we needed to remind Clark Gregg, “You’re not looking into Brett’s eyes. These dots here — above his eyebrows — that’s where his eyes are.†As he spoke, he was dotted up so we could catch all the nuances of Brett’s performance while synching it perfectly with all of his dialogue. Sometimes you get that Valley of the Dolls look, you know? You look into the character’s eyes and you don’t see a soul. What was important to me was to capture what Brett was doing — it had to be slightly exaggerated for Hive’s face, but it had to have a soul. You’ll notice, when he feels he has the upper hand, the animation is very languid and graceful on the tentacles. But the minute he starts to realize something is wrong, you see the tentacles stiffen. We spent a lot of time just getting the details right, so it would feel like you were looking at someone who was actually looking back at you.

Westworld Super Bowl trailer (2018), robotic bull stampede

Jay Worth, effects supervisor: The story for the bull sequence is, we’re working on the Super Bowl spot and there were no VFX planned. It was just gonna be some clean-up as the practical bulls ran through the practical set — the animal wranglers had these huge metal gates set up so that they wouldn’t be running through glass walls, they were able to train them to run this path, and then we put ‘em on the set and shot ‘em with a Phantom camera. All of a sudden, Jonah [Nolan] was like, “We have a problem.†They’d been doing testing, and the people who weren’t super familiar with the show, they weren’t clear that the bulls were robotic bulls. So it was a confusing visual. He’s like, “So we need to make it a robotic bull in three shots.†I’m looking at the calendar. I’m like, “Dude, it’s Friday night and this airs a week from Sunday.â€

All of a sudden, we’ve got six days to pull off the most scrutinized, public thing we’ve ever made. Thankfully, because of Dolores and all the other robots, we at least had an idea of how it should look, with our matte black finish and all that other stuff. But the team at DNEG was just amazing. We had a design within 36 hours. It just bounced around the globe — it went from London to Singapore to Los Angeles to London to Singapore, getting passed off around the clock, and it was really flawless and beautiful in a super short period of time. We kept joking, “Guys, this is a horrible precedent we’re setting, right now.†[Laughs.] The fact that we were able to do this is really not good in the long run. It’s great, but it’s not great.

Doomsday (2008), exploding rabbit

Neil Marshall, director: There was one particular effect that was a real nightmare — the exploding rabbit in Doomsday. There’s a scene where a machine gun shoots a bunny rabbit and it explodes into pieces, and it took us many attempts to get it to work. In the final film, it looks flawless. It really looks like we blew up a rabbit, and I’ve heard people complaining about that ever since, but it is an effect.

The first attempt was with a toy rabbit that somebody got in a shop, and it looked really bad. We tried so many times, we had people saying, “Why don’t we just shoot the real rabbit?†I was like “No, we’re not going to shoot the rabbit.†In the end, we had some rabbit fur and a sandwich baggie full of blood and blew it up. It boils down to using CG to pump together an exploding bag of guts, a real rabbit — we had a hell of a time just trying to get the rabbit to do what was necessary — and some machine gun fire. The visual effects people built a set with blue rocks that matched the other set, and then they blew up the bag of guts and scattered them so it’s a flawless, seamless joining.

It was a crazy, stupid difficult effect in a film full of crazy effects. We did a stunt roll in an armored car that was like 50 tons or something, but that was easy. The rabbit, that was hard. But when you’re watching the movie, you would really believe that we’re blowing up a bunny rabbit. It gets a laugh now because it looks so realistic. I didn’t want to get a laugh because it looks stupid. I didn’t want it to look like a rabbit from Monty Python and the Holy Grail.

Homeland, “The Tradition of Hospitality†(2015), IED explosion

Lesli Linka Glatter, director: Season five was shot in Germany, so when I got the second script and it had a whole huge section in the Syrian-Lebanese border, I thought it was a joke they were playing on me. But unfortunately, it wasn’t. We had to build a whole refugee camp and make it look as if it was on the border. I created this enormous set with NATO stations and tents and hard structures, and like a week before we started to shoot, the set had become so expensive that I had to cut it in half. It meant replanning and redesigning the whole thing, which I did. We added, with effects, thousands of tents in the background, so it felt like it went on for miles and miles and miles. And at the end of it was this huge explosion — Carrie notices that the camp has been evacuated, but she stops the car before the explosion goes off. There’s debris raining everywhere. The car, had it gone ten more feet, it would have exploded. The setup for that, in another language with a new crew, was, uh, funny and challenging and complicated. And that huge explosion, we were working with our German special effects team for the first time. It was quite a bit bigger than what we had talked about. We were all like, “Whoa! That’s dramatic. Maybe a little too dramatic.†So we ended up doing the explosion twice, and using parts of both explosions. It was all very challenging, but ultimately a powerful sequence.

Harry Potter and the Goblet of Fire (2005), Death Eaters

John Kilshaw, effects artist: The one that most required the greatest creative input was the Death Eaters from Harry Potter. That was a bunch of fluid simulations, which we used to drive a whole lot of cloth simulation. So you had simulation on top of simulation on top of simulation to create the overall effect. When we did the Death Eaters the first time, the software wasn’t big enough to be able to do what the client wanted. We had to create lots of small parts of the simulation that all work together. So it was like chaining together lots of different effects over a large area. The client wanted it as a smoky simulation – not to just get left behind – like it was smoke from a fire. They wanted it to move and evolve and travel with the character. So, you’re having to give something a life and a character and a movement all of its own, which is, again, not what the simulation software wants to do. We had to create software that would allow us to get in and get that out properly. Some bespoke tool sets were created to allow us to do that.

Bird Box (2018), rapids escape

Susanne Bier, director: The hardest special effect was the rapids with the kids and Sandy [Bullock], because it’s such a combination of Sandy and the two kids, stunts, effects. They’re escaping down the river, Sandy and the two kids are all blindfolded and they need to navigate the river blindfolded, falling out of the boat and into the rapids, they’re still blindfolded. They basically have to swim and not look, and they did it. Sandra Bullock was literally blindfolded the entire time and the kids were sort of semi-blindfolded.

We spent a lot of time with the stunt coordinator trying to figure out what can we do with the small kids. Rules are really strict about what we could do with kids, but also because in the actual river the water is very, very cold, even if we could do things according to whatever the rules are, you can’t do that to a small kid. For the tank to seem ferocious the way the river is, it needs to be relatively deep, so even in a tank there are limits. So they were thrown into the tank, but they were also taken out very quickly. So it was a very particular thing getting the whole thing to seem seamless and still be considerate of the kids being very small.

The whole point of that sequence is it needs to be real. This is not a sequence where you go, “Ooh, they did crazy effects.†This is a sequence where you go, “Oh my God, we have small kids turning around in crazy rapids in the river.†I think in terms of the logistics, in terms of actual effects, and in terms of real scare — because we were at the river with the kids, and they were in the water, and we were in the tank, and they’re really small — taking all the elements into consideration, that was the most daunting.

Miss Bala (in theaters February 1, 2019), altering an explosion

Catherine Hardwicke, director: Mine’s a little bit of a two-parter, because we had this one big, fun action scene, right in the middle of the movie. There are some clips in the trailer of it. We’re outside this big bullring, right on the border of Mexico in Tijuana. Gina Rodriguez is trying to escape from the cartel. She’s running, and the only way they could figure out how to escape was to blast through. They shoot a grenade launcher at these cop cars that are blocking the entrance. So that was real cop cars, on a real street, where there’s a lot of nervousness about violence already, in the middle of Mexico. We were really scared, too, because there was this beautiful art that’s been there for a long time, with the statue of the Virgin Mary. We didn’t know if they were putting too much explosives, and even the stunt coordinator goes, “I wouldn’t be shocked what they’re doing, if the whole thing is gonna go down. The art is going down.†I’m like, “Oh my God.†You know, this is a historic building.

But anyway, that went off great, okay. We have these really cool physical explosions that you see in the movie. And then when we were in post, we realized we had this one character that we actually wanted to die in that bullring that we hadn’t killed. So we were like, “Oh my God. How can we kill this guy so that his story arc ends there? It’ll be better for the movie. And we didn’t kill him in the scene.†[Laughs.] So we’re brainstorming. “How do you do it in post, in CGI?†And we found out that we could have him near one of the buildings that’s in this bullring. And then we would take another shot of the grenade launcher, flip it around, make it look like it’s aiming at that building, and then do a CGI explosion to blow up the building. Even though we never really blew it up, but it still had to match the lighting of all the other firebombs, and the other grenade launcher’s explosions at the time. So it was very fun to figure out how to kill a person that you didn’t kill. [Laughs.] How to stage it so it fit into a whole action scene that seemed really real and natural and organic, and match everything that we had already done.

The Hardest Effect I Ever Pulled Off