Trevor Hogg chats to production visual effects supervisor Jim Berney, visual effects producer Greg Baxter; and visual effects supervisors Matt Dessero, Marshall Krasser, and Berj Bannayan about bringing a dystopian Chicago to the big screen…
Greg Baxter
“I hadn’t work with Lionsgate or Summit before,” explains VFX Producer Greg Baxter (Jack the Giant Slayer) who was recruited to become a member of the Divergent production team led by filmmaker Neil Burger (The Illusionist). “There was a long stage of ‘Do you green light or do you not?’ By the time the button was hit they were racing to get it going. It all happened over the holidays with a bunch of phone calls where they were vetting me and I jumped right in. A lot of the crew had the same experience. I’m not the young adult target audience so I quickly read the books and script. I got to know who the characters were and tried to figure out what the visual effects aspect was going to be. Every movie is different. You have your horror and Sci-Fi films, and this was in-between the lines movie for us.” Over 1000 visual effects shots needed to be created for the cinematic adaptation. “For us, it was helping with the production design and helping to tell the tale of a somewhat futuristic Chicago but stay true to the elements of Chicago that were held over. In a photo-real world we had to match the buildings that were still there even though they were augmented for whatever the time and damage of the future held for the city.”
Jim BerneyJim Berney [Eagle Eye] was hired to supervise the visual effects. “This was my first show with him so I didn’t know Jim before the show,” remarks Baxter. “We both had to dive in and run with it. Jim is great; he has a long history in this business and had done a different kind of movie but the visual effects were fairly similar. Jim did I Am Legend [2007] where you had to show a real city [New York] that has been abandoned; that was another big challenge for us because in Divergent the population of Chicago is so small that most of what we shot on-location we had to remove all of the people, cars, and street lights. Jim set the mantra early by saying, ‘Never underestimate the city.’ Every time we were down there we would note all of thousands of things that needed to be replaced or removed and covered it the best we could when the camera was running and shooting so we could pull it off.” Berney laughs, “I knew he would say that! That statement got coined when I worked with Janek Sirrs [The Hunger Games: Catching Fire] on I Am Legend. He was the one who was pushing for more money. Usually you’re less money as you’re trying to get the job in. He’s like, ‘You have to put the money there. Don’t ever underestimate the city.’ As soon as Greg came on and we were looking at it he went, ‘You don’t need that much.’ I said, ‘Don’t underestimate the city.’”
“Downtown Chicago is a bustling city,” notes Greg Baxter. “There’s not a whole lot we can shutdown and shoot. Safety wise you can’t put actors hanging out of buildings 20 stories up. Once we locked down the locations and found our physical shooting limitations it became Visual Effects and Art Department figuring out what the stage layout would look like, how much green screen we would need for the camera coverage, what set pieces needed to be built and we ran with it. The train jumping on and off is a good example. We shot that four or five different ways. It had a lot of green screen and we had to rebuild the city in every direction. It was a big thing to do but was good in the end.”
Having the principle photography in Chicago was helpful to a degree. “If we went around and shot Chicago it would look like Chicago today,” notes Jim Berney. “There’s a war and this is after it. It’s not all of a sudden people are flying and there are hovercars. Its still technology a little bit further from today but it has been stopped because of whatever event has happened. That being said the city is still lived in so it’s not overgrown with weeds. They take care of it. Where the roads are they’re not using them for cars it’s more foot traffic. Eventually things like paint lines would go away, asphalt would erode down to gravel, street lights and street signs wouldn’t be necessary so they would repurpose them or melt them down because they wouldn’t have metal or the ability to go out and mine. We took it from there so a lot of the city work isn’t adding but taking away.”
Other key members of the production team were Production Designer Andy Nicholson (Gravity) and Cinematographer Alwin Küchler (Hanna). “Andy and Alwin early on had to figure out how to shoot the starting and ending point of each scene,” states Greg Baxter. “Alwin brought in a specific and artistic lighting look for the film. A lot of the discussions were the kinds of lights we were going to use and how the source in this world where there isn’t a lot of power and people was going to be. It had a significant impact in how we built and lit the sets and how we were going to continue that look in the CG world where the sets had to be extended. There was a lot of daily communication about that. Alwin would jump under the hood and do a quick colour correction and balance, and make sure Neil was happy with what should be the final look of that shot. We were able to use that concept to run with our future colour corrections; that was a good dynamic. In post with Film Editor Richard Francis-Bruce [The Shawshank Redemption] it was rebuilding the story. We had the screenplay, a plan for each scene but as we shot it and how the action unfolded everything got jumbled around a bit. There was a lot of rethinking about bridging one scene into the next. We ended up going back to do a lot aerial photography as transition shots.”
“We didn’t previs everything,” explains Greg Baxter. “We didn’t have the luxury of time for that. We used The Third Floor. Our trickiest scene was the Mirror Sequence which was a mind scrabble trying to figure out how it was going to look and back out of that how to shoot it. We ended up doing the whole thing on green screen with six or eight cameras per shot so it was one actress with a bunch of different cameras for her reflections put into CG mirror planes and built the whole virtual environment. We had to previs the heck out of that one. We techvis it to figure out where each camera goes in a physical space, how we’re going to build that set, all that green screen and how Alwin was going to light all of that without the lights getting in the way. It was craziness. We did previs all of the virtual environment scenes and the train work which was also tricky; that had to be carefully previs and techvis so everyday when we went to a location we knew where the camera was suppose to go, how much green screen was needed and in what direction, what the time of day would be and where the sun was going to be, and how that was going to affect everyone. There was a lot of work there.”
“When I first had the meeting with Neil, I went in and he had this piece of paper and explaining to me this Mirror Sequence and I got it,” recalls Jim Baxter. “Of course when I walked away from the interview I had no clue on how to do it! I didn’t have a clue for about three months. When I got to Chicago, Andy Nicholson had his guys make me a 14 by 14 inch mirror box with a little Shailene [Woodley]. There were people in it and I could remove the roof of it. I sat and stare at it at my desk. We approached it first story wise. I knew what Neil wanted with these different beats. I had to figure out what could we do. I used The Third Floor to do previs. Usually previs is just story and we did that on other sequences but this one I said, ‘I want to tech it. I have some ideas. I think for these different situations we could use different rigs and I want to explore them.’ We figured out the ways of doing these certain things and then we previs the story; that’s when we went back and forth with Neil, and that way we didn’t get painted into a corner where we couldn’t give him what he wanted. There were upwards to eight cameras but on average at least five for her. There were a 150 camera positions. We had to map them all out in advance. Originally, we only had two days to shoot but I talked them into two and a half days. It was a well thought out and executed sequence. It was textbook as far as getting the story out of the director, technically figuring out how to do it, getting everybody on board, testing it out and shooting it.”
“The Zip-line Sequence was a fun one too,” states Jim Berney. “It was a little nerve-racking. We put it off to the end because there was some helicopter work needed to done to grab the city. We had previs the sequence which everyone was in love with and I shot what we needed for it. We used mainly plates. I wanted to shoot everything. It is difficult to use plates for someone who was going to be hanging in front of a blue screen so it was going to be dangerous. But I knew how to execute without it looking ridiculous. I knew in a pinch we could go back to projection geometry work and we could rebuild camera moves. We put it all together. The studio after seeing the director’s cut wanted more. We went back and this was after shooting and revisited the sequence, re-previs and reshot Shailene Woodley. We went back to Chicago and grabbed some photogrammetry and some other plates. We gave it over to Scanline to do it all CG. We didn’t handcuff the DP or the director the second time; they did whatever they wanted to do and we put that together. It turned out pretty good.”
El trains feature prominently in the city. “It was one of my favourite moments on the movie,” recalls Berney. “It was a quiet time. Everybody else hadn’t gotten to Chicago. Andy had already been there for a month by himself; he was eager to get to work and figure stuff out with me. The two of us dug into that first. Like everything I do I don’t want to rely on CG work. I know it’s easier especially when you’re up against a deadline to blue screen it, get it to a vendor and do it later when you’re comfortable in L.A.. But I like to beat my head on it and try to get everything I can in-camera even if they’re pieces that you have to cobble together. The train is a good example of that. It came out fairly seamless but there were four beats and they were all shot in different places in different ways. It starts out in a real location in Chicago where Tris [Shailene Woodley] starts to climb up on the El; she can only go so high because we can’t get that close to the electricity because they cannot shutdown the El. We go to this other street which does not have a train line and built a set which is about a 50 foot piece of track where they can climb up and get on top of that. You can look both ways. We threw a green screen behind whenever we were looking down the streets. We had to digitally extend those, the track and the train. As the kids start running to catch the train they can’t be running up that high because it’s too dangerous. Then we built one car and a set of tracks in an old warehouse district. For them running along the train that was shot there because they were on the ground running. They got the train out of it and the tracks at their feet. It gave them something physical to run along and react with, you get the dust, the wind and they get to climb up on it. The rest of it we had to replace the whole environment and add the other three cars. It was fun to see it all come together. Method did a spectacular job.”
“They were a hodgepodge of ideas,” remarks Jim Berney when discussing the drug-induced Fear Simulations. “You go into the fear landscape by yourself. You go into your own head. There’s a beat where Tris and Four [Theo James] go into his fears which are height, closed spaces, killing an innocent and his dad. It starts off with height so they’re up on top of one of these cables which are strung between buildings. That environment is one of the few all CG because that location would have been impossible to shoot. We LiDAR scanned that aspect of the city as well as everywhere around there. They go into a hole of a building and it’s a confined space of a small metal box that is shrinking which is all practical. They come out in this really beautiful one shot where you see them then the camera swings around to a gun. Four picks up the gun and you swing onto him. All of a sudden you see an innocent girl sitting in a chair and swing onto Tris who is at a different location. We come back around his head and the innocent girl becomes Tris; he pulls the trigger and then she is on the ground but you don’t really see her. Four turns around and walks into a house where Tris is there and he meets his fourth fear which is his dad. There are five Marcus surrounding him. It’s all camera work and compositing. That one flows nicely.”
Theo James on-set with Neil Burger“It is creepy through editing,” observes Berney. “It is not through weird effects. When Four punches Marcus he falls down. Marcus doesn’t turn to ash and flutter away as a butterfly. None of it is in your face effects. When we go from the innocent girl to Tris that’s a compositing thing and when you see multiples of Marcus that was multiple passes of Ray [Stevenson], the actor who plays him, and compositing. When Tris goes into her first fear which is being outside the fence she’s on the outside of the fence and nobody knows what’s out there. We start with a flock of birds coming towards Tris so her fear of birds starts to chase her down and she seeks the safety of the fence. The birds are CG. Tris gets stuck in a practical mud puddle. Eventually, she gets down and is trapped by all of the birds and there’s a puddle of water where she sees her reflection and realizes it is not real. A lot of that is environment work as far as putting in the fence. Even the big fence that surrounds the city we were lucky enough to find a location that had a huge concrete wall. We could marry to it. The fence is over a 100 foot high and that is all CG except for the base. We get to tie into the base, the dirt and weeds around it. To extend this environment down to infinity we had to get rid of the lake or any cars that were way back there or any signs of houses and factories. There’s a big tanker back in that area which is suppose to be a dried up lake bed.”
“The Leap into Dauntless was another one of those well thought and planned out things where we worked with Neil with what he wants and picked the brain of Garrett Warren [Ender’s Game], our stunt coordinator. Garrett grabbed his team and did some camera work to give the editor something to play with. In parallel we were working with The Third Floor on the previs style the way the CG people do it, the stunt people were doing their thing, and we came up with some ideas. We didn’t want it be a super multi-cut thing that is extended too long but we didn’t want it to feel like she jumped into a net. We found that location on the roof. We had a practical area to tie it into. There was a big net down there for Alicia Vela-Bailey [Interstellar] to jump into. We did a face and head replacement and replaced the roof and put the hole in. When we’re down below looking up it’s a different location of the same building but at least the architecture was all the same. We had to remove the skylight and extend the building as a whole up.”
“There are a lot of great companies that can do this kind of work,” notes Greg Baxter. “Jim definitely had some experience with a few. Unfortunately with a lot of what is going on in our industry, people move around from facility to facility quickly so it was both vendors and their history but also the key artists within those vendors. Jim early on locked into Method Studios and Matt Dessero because he had worked with Matt before back in Sony. Matt has a good eye for lighting and the biggest trick for us on all of this was keeping everything photo-real and that’s a delicate balance. The difference between something jumping out as being CG and photo-real comes down to how you lit it to a background plate. We put a lot of work with Method in with their L.A. team led by Matt Dessero. Scanline I had previously looked at for a different film for CG city work and we vetted them and put their Vancouver team on Zip-line and the Four Fears High Wire virtual environment.” Soho located in Toronto handled the Choosing Ceremony and the graphics for the monitors. “Of course there was a financial aspect to it as well but we started creative and backed out of that to how we could put the pieces together to fit the budget.”
Shailene Woodley on-set with Neil Burger.“I’ve worked with Jim in the past,” explains Method Studios VFX Supervisor Matt Dessero. “For this movie we went through all of the artwork, and all of his ideas and plans as to how he saw things being shot. Jim wanted to capture as much in-camera as possible which I agree with. There’s no reason to do it all as visual effects. It always looks better when we have practical effects in there and as many set environment pieces as we can. That’s what Neil Burger wanted also; he wanted it to look photo-real. The goal was to make it look like a movie set in the future but believable. At the end of the day it was an invisible visual effects movie. That was the M.O. going in from Jim and Neil. We were game for that.” Dessero remarks, “By the time I got on Jim Berney and Andy Nicholson already had the established the lay of the land and had beautiful concepts for every scene. We had these gigantic turbine fans that encompassed the city and had some cabling throughout connecting all of these future power generators. It was mostly there in the concepts. We were there to execute and added our subtle touches. Everything was there from Neil, Andy and Jim. It was about getting this thing executed and figuring out how to plan some of these complex shoots we had to deal with.”
“I was on and off for three months out there,” states Matt Dessero who had an on-set presence during the principle photography in Chicago. “To help support the team we were sometimes two units and one point we were separate units but for the most part Jim had his data wrangler Alistair Jamieson and I brought Fabio Zapata from our end who is also our tracking, integration and layout supervisor. Fabio was there with me making sure that all of the footage and data we were collecting would fit right back into our pipeline back here at Method. From the Method perspective it’s awesome because we’re acquiring the data exactly as we want it. Fabio was an instrumental part of that and he was there with me ensuring that zero shortcuts were taken. He’s a meticulous guy. It was great to have him there. There were times him and I would say, ‘This is going to take a bit of time. Jim says we have five minutes to acquire these people.’ Fabio and I would discuss and come up with the shortest possible time to photo acquire somebody, optimize that process and run through it. We have worked together for years so it was easy for us to communicate what we needed and what was important to the both of us.”
“In one specific shot the Dauntless Army is about to storm the Abnegation neighbourhood,” states Matt Dessero. “There were 150 actors that we shot on-location and they comprised the foreground portion of the army. Our job was to extend that to 500 people. We had 350 digital doubles that were behind this army. The good thing was that we shot reference of the army. We brought that footage with us to the motion-capture stage and played it for our actors. We had the props there. They’re holding guns or a rifle. We posted it through the VTR to make sure that the performances were matching the speed and cadence of the plate photography. Months prior to this we photo acquired 20 different actors because we knew that we were going to build an army. Frank Zapata, our data wrangler, and I took visual stills of each actor from a wide full body and we would punch in from the mid-chest up for a close-up for higher resolution skin texture. Before the motion-capture shoot we were already processing these digital doubles. Once we captured the motion-capture performance we married the two together and had a digital double with animation. The digital double was rigged so we could animate them individually in our traditional animation setup before we would import that rig to the Massive pipeline. From there we would run them through the crowd simulations. The data was exported from Massive to Maya and we rendered Maya in Vray. All of crowd was lit the same way that our environments were lit.”
“The Abnegation Neighbourhood was built as a set piece on-location in a field,” remarks Matt Dessero. “There were 12 facades. Full houses, half façades and single façades were built. The environment needed to feel as if there were a hundred of these homes. It’s a big Abnegation faction. They’re like a project home, simple and all the same. We shot with green screens at the end of the façades and that allowed us to extend out to any direction. It was nice to have the pieces built because anytime the actors were acting they were acting in this world. We not only had to build the neighbourhood but build grass. If you drop a CG building on top of a grass field you have to add the grass that breaks up the edge. We also had to make it feel lived in. Logs and paths were added, and CG people were added off in the distance. If you extend these buildings out lets say 20 buildings in one direction and if there’s no people back there it feels vacant and boring.” Dessero notes, “Just having laundry blowing in the wind which nobody is going to know if it’s CG or not. It gives it a lived in feel. It’s the subtle cracking, weathering, aging, and repairs that have been done to the buildings which make it feel like a war torn land.”
“When Jim brought us on the Mirror Room was the first sequence that I saw,” recalls Matt Dessero. “That was the one which sold me on the movie. ‘That’s amazing.’ Besides the concepts it was a challenging sequence. Our first meeting with Jim about a year ago he discussed the Mirror Room at length. It was like, ‘I have to get my head wrapped around this, Jim!’ It took us time. There was a lot of back and forth with myself and my team. We took the previs that Jim and The Third Floor did and turned it into techvis which is a lot more exacting. Fabio and I went through that. Our other visual effects supervisor on the show Marian Spades spent a lot of time on it also. We broke it all down, we knew that we would be shooting with Alexa cameras and requested six Alexa 4:3 cameras for the sequence; they would give us more headroom as suppose to the Alexa 16:9 cameras. We readjusted all of the techvis to fit that film format and that allowed us to change some positions, move some cameras in or out, and lose a camera in some situations. We spent at least three months planning it. Jim made a test shoot happen. Everybody knew it was going to be difficult so we planned for a four shot shoot. We planned them all out, got the cameras, shot everything, and tested a couple of the camera rigs. One of the big issues was eye line for Tris because she’s acting with herself. We had to have our VTR setup so we could review the hero camera and the hero reflection cameras, and overlay those images. The VTR had to give us feeds from the other four reflection cameras so we could review and compare against our techvis and shooting schematics. It was logistically a difficult setup.”
“The reason why we wanted a common origin was we didn’t want to move the camera setups every single shot,” explains Matt Dessero. “We wanted to keep the origin consistent so that allowed our cameras to move around 20 feet in any direction. They didn’t have to move 180 degrees around the scene. On the floor of the room we marked off the actual size. Everybody from the AD, cameramen, DP and gaffers knew where the Mirror Room was going to reside. Now that you have that common origin everybody can start working to that origin. The techvis is for simplifying the shoot. We’re trying to simplify the shoot so it makes the most sense. You’re going into a big green screen stage which is 100 by 80 feet and you’re plotting the camera positions and having the cameramen setup and the target is this actress who’s floating in this green world. It’s an abstract way of shooting a scene. After we got through the techvis and all of these nice Maya scenes were locked down we exported the camera positions to our survey head and from there we would go back out to the stage and plot the camera positions directly on the ground. It’s robotic like a survey head. It would calibrate the room and take the 3D world and put it back into the real world.”
“It is mostly flat lit in that environment,” reveals Matt Dessero. “The DP Alwin Küchler and our gaffer Len Levine, Jim and I spent a lot of time trying to figure out during the test shoot how they wanted to arrive with the lighting. Jim’s and mine concern was you can’t just move lights around because they’re going to be seen or be visible by all of the cameras. The entire lighting rig was overhead on a gridded system. Jimmy Shelton [Key Grip] and his team built a gigantic grid overhead with light panels that could move in any direction; they could light the actress the way they best saw fit with a shot. None of the lights had bases connected to the floor.” Dessero remarks, “We had three computers standing around with the previs locked down for every setup. We would show that to Neil Burger, Jim Berney and Shailene Woodley. The first AD Artist Robson and all of us would look at the shot and familiarize ourselves with it. My team would be out there making sure that all of those cameras were being positioned. The stand-in was moving from the start mark to the end mark. We were framing all of that up with Trevor Loomis [The Lone Ranger] and Dave Thompson [Silver Linings Playbook] who were the cameramen. We had a guy on my team filming an extremely wide 10mm witness camera of the whole scene. We had another visual effects TA running blue plex and shooting HDRIs every setup. I had Fabio surveying the cameras at the end of every shot.”
“We had one shot at the end of the sequence where the character Tris is tackling a dog,” states Matt Dessero. “There was no dog there. We had a stuffy that the stunt girl Alicia dove into and grabbed; that green stuffy was replaced with a CG dog. Since she was acting with this dog that was on-set we photo acquired it. We had a five camera setup that they all triggered at the same time. Dogs are sensitive to movement so we didn’t want to move a single camera.” The entire environment is CG. “The ceiling and floor were all based when Tris first enters this Mirror Room which is all happening in a simulation. Tris is being tested to see what faction she should go to. Tris is in chair and they hook her up to a device and she mind wise enters into a Mirror Room. The lab that they begin in is an actual set piece. Tris is in the chair and there’s a single mirror plane on one wall. The rest of the room has a floor and a ceiling and that’s what everything was predicated from. We needed to extend it all out. If it were a perfect Mirror Room there would be straight lines receding off into the distance which is rather boring and looks CG. We added rotations to our mirrors which gave us this banana effect. We added the curvature, did some tests and delivered those to Neil and Jim; they picked the right amount of curvature for them and we went from there.”
“In the opening scene of the city the camera starts out in Lake Michigan,” states Matt Dessero. “There’s not much water in the lake as it’s mostly swampland. Jim shot some beautiful helicopter plates. The water was replaced with our terrain. The terrain was a combination of procedural terrain, a generation system we created for this show and projected matte paintings for the buildings destruction. There were also trees, water and birds out there in the distance.” CG trees were incorporated into the scene. “We started with a tree base and leaves and added some springiness to the branches. In our most complex version of the setup individual leaves moved also.” The setting required more vegetation. “The camera booms up and starts to rotate around. Eventually the camera stops because there was nothing else out there. It was dried land. We needed to extend that into a full 180 degree turn. We tracked the shot and took over with our CG camera. We added another 70 degrees of rotation. We had to create all of the grass that was seen in the beginning of the shot. We had a grass system that included grass, grass seedlings, and dew on the grass. The grass was right to camera so we had to blend 100 per cent perfectly to plate. Our CG grass had a wind simulation that ran through it to give the same amount of movement as the original plate grass had.”
A massive structure encircles the city. “For one of the scenes we shot in a location that had this base of a fence,” states Matt Dessero. “I was there with Fabio Zapata. We surveyed, LiDAR scanned a chunk, and built a CG fence. We knew that we were going to have to add the towers which were concept designed but stood another 100 feet tall on top of the fence. We added some cabling and tension wire to those towers. There was a set piece for the guard shack that we had for one of the scenes which we also photo surveyed, LiDAR scanned those pieces, built them in CG and extended them out. Anytime you see that fence outside of the Go-to-Fence sequence it’s all CG. Wind turbines also appear. There are about 850 in the wide shot of Chicago and cables are draped throughout the city. We ran them all through animation or simulation depending on how and what we wanted. Each turbine created looked similar to a jet turbine. The movie takes place a 100 years in the future so wind turbine technology has progressed through three generations of turbines. There were the simple clunky ones then we got to the mid-future century one and moved into a modern day. They all have nine extra variations on each of them; that’s different amounts of staining, weathering, grime and grease. We would dress those accordingly into the different shots. If we were in a more rustic or decayed part of the world we would use our more dirty turbines in that environment.”
“They built one side of a single train car with doors that opened and closed, and the interior was built and dressed,” states Matt Dessero. “It was beautiful. The back half was not built, and the front and back of the train were half built. They built the inner doors but there was no roof on the practical train. We LiDAR scanned, photo surveyed and photogrammetry it, and recreated it in CG. Every time you see the train there are four cars connected. We put on some variations of dirt and rust maps, extended the train out to a four car system, and rigged it so that each individual car had a separate rock and roll control. Each of them rolled against each other a little bit as they went down the track. We had to design the top of the train because there was no roof. We also had to build all of the trusses and tracks that the train is riding on.” The Dauntless Compound was a big build for Method Studios which was inspired by a former military depot turned into a manufacturing warehouse. “We photo acquired six different buildings to get the different brick and window patterns and arches that we saw. They were all similar buildings but had subtle differences in the aging, weathering, and glassworks that were on them. We also had them LiDAR scanned. The six buildings stretched for a mile and what we built in CG was about two to three miles worth of buildings.” A glass atrium was constructed. “The opening shot of the Dauntless rooftop is two green screen plates that are merged together to form one single plate. There’s a chunk in the middle where we reveal our CG atrium. There’s everything on there from bird poop to a collection of aged dirt and weathering on this glass to make it look photo-real.”
“We shot people jumping from the train to the building in some scenarios for the medium and close-up shots and some of the wide ones also,” explains Matt Dessero. “It all depended on the angle. Jim’s preference was always to shoot it in-camera which I agreed 100 per cent. We would shoot the stunt performers. Anytime there was a shoot with people jumping Fabio and I would have multiple witness cameras running and that allowed us to roto matte their actions. We had a good base for our animators to work from. Also during our motion-capture shoot we had all of the motion-capture actors run and jump as if they were jumping onto the rooftop from a train.” Complicating matters was the fact that the actors and the digital doubles were jumping from a moving surface. “You have the inherent velocity and it’s coming from the train so they’re not jumping in a straight line across a path. They’re jumping and travelling at whatever their inherent velocity is. In this scenario it was roughly 10 to 11 miles per hour. In some cases we sped the train up to 18 miles per hour to make it feel like it was moving faster. We would have them jump from the train, take the motion-capture performance, add the train momentum and capture that inherent velocity through the shot.”
A flock of ravens attack Tris during her fear simulation. “Yves DeBono [Due Date], the special effects supervisor, had little air guns that would blast air at Shailene Woodley, disrupt her hair and try to shock her. They were loud little devices and blast a lot of air which was nice because Jim’s and mine concern planning this was how are we going to get that interaction with her? Originally, we were only going to have 15 birds in these scenes. Ultimately, we ended up adding 150 birds in the biggest scene. The good thing is when we were out there shooting Fabio was with me. I said, ‘Let’s make sure we witnessed cameras these so we can try to triangulate where she is in 3D space.’ I had a feeling we were going to add more birds.” The scene echoes the horror classic The Birds (1963) directed by Alfred Hitchcock. “It’s very reminiscent and one of mine favourite movies. We added scratches to her face in post so we could align our swoops and scratches with the scratches that were on her face. We timed them all accordingly.”
“We have built a lot of birds here at Method for the last few years so we have a decent system for it,” explains Matt Dessero. “We took it one step further on this show. We started building the birds at a higher resolution. I always build things at a higher resolution than I think we need because I don’t want to have to go back and add the detail later. Once we had this many birds we had to dumb down our rig and create a low resolution version for the animators to be able to animate the scene.” The background birds were simulated while hand animation was used for the attacking hero birds. “The details on the claws were Zbrushed in, and displacements maps were created. We modelled the face, eyes, and feathers. Feathers were applied in Houdini through our effects feathers system that we built here. I wanted curvature on the feathers and wings; they could puff up or bend or curve. We had moulting feathers under the larger feathers. We went through that entire phase of look development getting the reflections and iridescence right.”
Tris gets trapped in a water tank. “That was beautifully shot by Alwin, the stunt team and Jim,” remarks Matt Dessero. “It was simple and straightforward. We added the glass crack when she’s tapping. In the wide shot we added more fissures to the glass to tie-in with the close-up cracks but they actually blew the glass and the stunt girl came out of the tank.” Conventional bullets were not fired. “It was based on a traditional muzzle flash but they wanted a blue tinge to them. There’s a double compression band that runs around these guns.” CG assets were shared with Scanline such as the wind turbines. “We rendered some terrain for them for a couple of shots of the looking out at what used to be the lake. We shared our digital doubles and gave them the simple model and the textures.” As for the biggest challenge, Dessero states, “Without a question the Mirror Room. We planned that thing for three months, Jim, me, Fabio, Marion and the team back here. The second biggest challenge was building the Dauntless environment. Those brick buildings were gigantic. The glass atrium was a big deal. It was a huge challenge to make that environment look photo-real. I believe we arrived there. Neil, Jim, and everybody seemed to be happy with those environments.”
“Jim Berney and I developed a rapport while on set; we talked long and hard about what his and Neil’s expectations were for the show,” explains Scanline VFX Supervisor Marshall Krasser. “While we were in production, this understanding allowed us to be on the same page and work toward the common goal. Jim would always keep us grounded and made sure that what we did was reality based and not ‘science fiction’ futuristic effects.” The emphasis was on producing invisible effects. Having worked on a number of past projects that took this approach, we were able to guide our team to deliver these types of ‘invisible effects’. The key is to understand what reality is, and when we say reality we mean the reality that the camera and cinematographer has captured in the non-VFX shots. The last thing you want to have are shots that have deviated from the look of the rest of the movie. Being on-location is critical, since you can observe and commit to memory the approach that is being used by the Director and Cinematographer. Were they using atmospherics? Adding cucolorises or other gobos? Keen on-set observations coupled with reality based references are the essential components of successful ‘invisible effects.’”
Marshall KrasserA balance needed to be achieved combining futuristic elements with relatable ones. “Jim and Neil had a good grasp of the needed blend between these two needs,” states Krasser. For the Capture the Flag Sequence, we were asked to modify the flag that was used to make it more futuristic, yet grounded in reality. We researched possible references and came across a relatively new fibre optic cloth that is on the market. Andy Lesniak [CG Supervisor] ordered it and wired up a system that would allow us to pulse light in multiple colours through this material. That was then shot and used as a base for our effect. For the ‘tracers’ in this sequence we wanted to avoid the look and feel of typical science fiction tracers. For this, we looked at footage of an archery product that is a LED arrow nock. This look was a little too extreme for us, so we dialled it back to a more subtle and suitable level.” “We had several environments that were pretty extensive and complex,” remarks Marshall Krasser. “The Pipe Bridge Sequence that you see in Four’s Fear Dream was a very fun and challenging build. The very ornate and complex Jeweller’s building was a satisfying and rewarding challenge. We shot still photography from multiple vantage points, this gave us a great base from which to build and texture something that was going to be showcased front and centre in several shots. The unique challenge was that it also needed to seamlessly blend into the small section of set piece they built for the green screen shoot. The set designers copied the original pretty accurately, so there were only slight modifications that we needed to implement. In this sequence, we also built the Trump Tower, and Kemper Building, plus dried up the Chicago River. For the more distant buildings that did not require moving reflections, we utilized high resolution still photography that was shot near the location for projection maps. For the final kiss of realism, we added atmospherics, cloud shadows, flying birds and insects plus blowing leaves and particulate.”
“In the Zip-line Sequence, the plan was to utilize as much of the aerial helicopter footage that we could,” reveals Krasser. “As the sequence evolved, we found it necessary to digitally re-build most of downtown Chicago along the zip-line route. Since we had shot extensive aerial still photography of the Downtown areas, we were able to build a detailed and precise model of the required path. Also, as any native Chicagoan has figured out – the beginning of the Zip-line and the end alley locations are not in a direct line. In fact, a 90 degree turn would be needed to link the two locations. By building the city asset as we did, we were able to re-align our run to make all of the plates and locations work. To augment and tell the story, we also modelled numerous buildings that we interacted with, like the ‘Fan’, ‘Mirror’, ‘Hole’, ‘Ski Lift’ and other key buildings. We also created a ‘Ferris Wheel’ model that was used in this sequence and the Climbing the Ferris Wheel sequence. Our initial plan was to only create the top of the Hancock building, but we had a particularly difficult helicopter fly over plate that did not want to cooperate with our needs. We felt the most efficient approach was to re-create the entire move and environment digitally, and it worked out great. This approach allowed us much more creative flexibility and control.”
“For the Zip-line Sequence, Andy [Lesniak] and I took a trip to the local mountain zip-lines and shot reference footage from six different runs to experience the dynamics involved,” explains Marshall Krasser. “This type of real life experience is something that you cannot always take advantage of, but is very helpful when you need to portray something that is very first person in action. Our goal was to involve the audience and make them participants rather than observers. For the Four Fear Pipe Walk, we spent time on top of these and other buildings doing visual research. There again, the key is observation and evaluation of similar environments to acquire a ‘feel’ for the reality from this height. You much constantly challenge your interpretation of reality and how you can stylize this.” Krasser remarks, “We were called upon to help expand the scope and scale of the Dauntless’s Pit and Chasm throughout the movie. Using the set as a guide and other client supplied concept and photography we were able to build an extensive digital model that allowed us to move freely around the environment. There were several digital double shots scattered among the sequences. In the Four’s Fear Sequence, we utilized our motion capture stage remotely to direct our capture supervisor in L.A. while we viewed [in shot] his performance in real-time on our desktop machine in Vancouver. This approach allows for instant evaluation and allows one to direct the action from almost anywhere in the world. We also used Massive to handle the larger crowd simulations we delivered. Additionally we helped Tris and Four climb Navy Pier’s Ferris Wheel and augmented the subsequent capture the flag battle.
Practical and the visual effects need to interact with each other. “The best way is extensive pre-production planning,” observes Marshall Krasser. “That way, on the day you are shooting, all parties are aware of the needs and expectations of the other. Do you need on-set atmospherics or will you add that in post? Do you need wind, if so how much? What is the desired lighting and does it require special on-set attention? In a nutshell, it’s asking the right questions. After that, it’s a matter of following through with the plan and hopefully you have covered all bases. Previs and concept art are excellent tools to guide everyone through this process and is the most cost effective approach.” The biggest challenge was the Zip-line Sequence. “After looking at our first tests, the studio decided that they wanted to expand and refine this sequence during post- production. This new cut almost doubled our shots just prior to the end of the year when only six weeks post were left. We had to add several full CG buildings to our Chicago model and expand the environment.” The end result worked well. “We were very happy with the way the Zip-line and Four Fears Sequences came together and flowed in the movie.” Krasser adds, “Divergent was an exciting project for us at Scanline; it was a film that allowed us to showcase the diversity of imagery we create and work with a great studio side team.”
We did the Choosing Hall Sequence when the kids are choosing which faction they’re going to go into,” states Soho VFX Co-founder and Visual Effects Supervisor Berj Bannayan. “In the five factions they had a hundred extras for each which filled the church as it existed. We put in another 2500.” Full 3D models were created for the digital stand-ins. “For each faction we built a variety of men and women with a different selection of clothing, hairstyle and skin colour. We have our own crowd duplication software whereby we can define a certain animation cycle for each person so they’re shuffling in their seat or clapping or walking. We had to also to do the filling in of the Choosing Hall as they walking and shuffling past people sitting down. We had a software tool that the animators could control in a certain way to make a random distribution of men and women with a variety of clothing so that they didn’t look like they were copied and pasted throughout.” Bannayan remarks, “The production took photographs of 60 to 70 people standing and sitting in a light tent so there’s an even lighting. We used that to build the models.”
“We went through an approval process with the director about how many seats we would add and what the church building would look like because it was completely artificial,” explains Berj Bannayan. “Originally, we decided to double the number of rows of people and Neil wanted a grander vision. It ended up being quite a lot larger than that. Once we had laid out the new Choosing Hall with all of the seats we were able to use the crowd software to put a digital person in each seat. The church itself was a semicircle so it worked out well in terms of the story that there were five factions and there were five seating groups already in the church going away from a central stage or altar. Each of those five sections were already in a pie shape so we extended each of the aisles that were between the groups of people all the way and filled in each wedge with more seats. It leant itself to being expanded that way so production chose well for that. The ceiling was much lower than it needed to be so there were some columns in the back holding up some offices above the church and they were 10 feet tall. We made those columns in the digital church 20 or 30 feet tall to accommodate the larger environment.”
“We built an entire church with the parts that were real on-camera but would only render the areas that were fully digital,” states Berj Bannayan. “In order to allow us to track the camera and make sure that everything was lined up properly we built the real church as a fully digital model.” A LiDAR scanner was utilized to assist with digital construction. “The digital cameras were set by the practical photography and that allowed us to augment the environment. The director and the director of photography were careful to light it the way they wanted; they wanted the particular colours to be the way that they were. There were these particular pools of lighting that they had over different crowds. That reality was set on the day while we were shooting and it’s important to respect the look and vision that they have.” Bannayan notes, “When the lighting bounces off the people sitting in the chairs the colour of the clothing casts some colour onto the marble that is surrounding them. All of that is part of believing that illusion that it’s real. It’s something, we get quite good at over the years.”
“The crowd and building were all rendered at the same time,” remarks Berj Bannayan. “We have come from a tradition in visual effects over the last 20 years where you do a lot of cheating. ‘I’m going to do a bit of colour correction here and there.’ Because the rendering wasn’t quite as advanced as it is today. But now that we have a lot of computing and processing power, the software is a lot more sophisticated, and even as artists we’re much more sophisticated, we can light and build them for real. If there was a light above them which is a particular colour temperature and a particular size we can put that particular light in a digital world. By doing that it makes the job much simpler to make sure that everything will match. If we took those people and put them outside we would light them with outside lighting and they would light a person who was standing there outside. That worked out well for us. There was a certain amount of artistry to it. That’s what the compositors and lighters do. It’s not all mechanical work. There’s a certain flare and artistry to knowing when and how to match to the real world.”
“The biggest challenge for the Choosing Hall Sequence was the volume of the crowd and getting the movement,” notes Berj Bannayan. “In the story they’re essentially sitting in church; that was the mood that was set for us. They’re respectfully sitting in place so there are lot of subtle movements. People can’t be perfectly still but at the same time you don’t want them to be moving around so much to call attention to themselves. Each person had a fully articulate outfit that they were wearing, some cloth simulation to make sure that everything when they moved flowed smoothly.” Bannayan states, “We didn’t get super close to the people but they all needed to have articulated faces. You can definitely see that they have faces and digital fingers in some cases. The animation needed to respect that. Their hair needed to move in some cases. It wasn’t like a full screen tight shot on a person level of detail but at the same time they weren’t little stick figures. It was that balance where in some shots they were tiny people and in the next shot we got close to that group of people. We had to build the character for both of those situations.”
“The monitor graphics were an interesting challenge because in some cases they were pure scenery,” observes Berj Bannayan. “It’s something in the background taking up space and giving some texture to the background. In the majority of cases people were interacting with them. The monitors were all touch screens. They were moving things around on screen and we had to make it feel like it was a real thing. We had some designs that came from production that we used as a jumping off point. There was a cobbled together feel for the technology. There was a blurry and burnt out edge on the monitors. The monitors have a bit of noise to them, some texture and a bit of flicker. All of those things make it feel like it wasn’t a perfectly pristine hermetic environment that they were in. It was a scavenged environment.” Bannayan states, “Aside from the mechanics we did 160 to 180 shots with monitors. There’s the mechanical work of the green screens and tracking in the digital monitors and making sure that everything is in the right place and the compositing and roto are all accurate. Those parts are you work on them as you go and they’re a lot of work. You want them to feel like a real thing. It doesn’t look like Minority Report [2002] or Star Wars [1977]. It wants to be its own world. At the same time it can’t take you out of the movie. We went through a number of iterations. What does the monitor physically look like? Does it emit light or is it more like a Kindle where it’s a surface that reflects light of various colours. Is there a bit of a glow? How much does it flicker? What are the different colours? What is the colour pallet of this universe that we’re in? It was an interesting process.”
“Getting the look was a challenge,” notes Berj Bannayan. “There were the different kinds of monitors to keep track of. There were surveillance monitors that you would see from a CCTV so they had a particular kind of video look they wanted with some readouts. There was also the mind’s eye when you’re looking through someone’s point of view during the fear landscape. At the fight in the end you can see Four’s view through the monitor over the top of his head. The mind’s eye look had its own iterations on how recognizable was it, is it too noisy, or is it clean? While we were looping through all of that design process the bulk of the compositors were perfecting the shots. When the look of a monitor in a sequence was approved we finished all of those shots. The other challenge was volume. At the end of the January we were down in L.A. for a couple of reshoots for the finale, the fight between Tris and Jeanine [Kate Winslet] and Four. We had that environment there with lots of monitors all around her and she’s interacting with the countdown in the red squares. We received those elements. That particular design was done by another house for those particular monitors. The studio had someone do that particular design so we had to key all of those things. We had 60 shots that we had to do in two or three weeks. It was a mad dash in the end because they wanted to redo some of the story at the end so there were tons of monitors.”
“We did have a couple [of shared shots with other VFX vendors] in Tris’ fear landscape,” states Berj Bannayan. There’s a section where she is lying in this orange chair and everybody is standing around looking at the monitors of what she’s seeing so the footage that was in the monitor was supplied to us. We would put our monitor look and the mind’s eye look on top of that. That was about the extent of the shared work for us.” Divergent was a big project for Soho VFX which handled 200 to 300 shots. “Sometimes everything is handed to us and this is the element you’re going to put in the shot. This one afforded us the opportunity to be involved with the design process which was nice.” The Choosing Hall was the biggest challenge. “It’s about making something that nobody would even notice that you did. It’s not like we put a million people in there. There are 3000 seat theatres to be had in the world. The goal with that shot was if you asked somebody what visual effects there were they would say none. I always want to make it feel like that we didn’t do anything at all. I feel like we succeeded in that way.”
As a standalone sequence the Mirror Room was the biggest challenge. “It’s creepy and really interesting,” states Jim Berney. “I don’t know if people are going to pickup on the subtleties but when Tris wakes up the Abnegation don’t look at themselves in the mirror. All of the sudden there’s the one mirror on the wall and she turns and looks at it. Even that is all green screen we had to put together. You get up with her, go to the mirror, she looks to her left and now you’re looking onto one of the walls. As you do that pan the two side walls become mirrors and you don’t notice that happening. All of a sudden she is reflected into infinity and you’re like, ‘What!?!’ As she looks back at her original self the back wall has now become a mirror and now it’s gone everywhere. It cuts and you see her walking away from herself and at some point she comes to herself. You’ll understand it as a mirror because you can see little bits of glint on the edges. Tris walks around herself. It’s as if the geography of the room has been mirrored and multiplied.”
Greg Baxter agrees with his colleague. “Shooting wise our trickiest scene was the Mirror Scene because there was such a complicated tech setup and execution of it. The other big picture issue was anytime there’s a first movie of what is hopefully a successful franchise the first tends to be rather limited in budget.” The VFX Producer also enjoyed the Zip-line Sequence. “Tris does her first zip-line through the city and it’s her first ‘I Am Dauntless Moment.’ When we shot it the first time around our schedule had been so tight that we only had half a day to pull it off. It didn’t seem like it would be too difficult because it was one person on a wire on a green screen. It was difficult and some of the shots we rushed through didn’t quite pan out so the first blocking of it in post wasn’t as exciting as we wanted it to be. We were fortunate enough in one of our pickup days to get Shailene Woodley back on a wire. We had previs more action out of it to beef up the sequence so by the time that finally came together late in post because we had added all of these shots the scene came together well. It was the right balance of physical wind on her to sell that she was moving quickly, not too frenetic so you could ride the story with her and not be zipping around trying to figure out where we are and then Scanline pulling off a nice looking city that is lit by the moon.” Reflecting on the project Baxter remarks, “The challenge with this one was staying as invisible as possible and helping to tell the story without making it look like a big visual effects movie.”
Many thanks to Jim Berney, Greg Baxter, Matt Dessero, Marshall Krasser and Berj Bannayan for taking the time to be interviewed.
Make sure to visit the official websites for Divergent, Method Studios, Scanline VFX, Soho VFX and The Third Floor.
Trevor Hogg is a freelance video editor and writer who currently resides in Canada.