With the latest Marvel blockbuster, Captain America: The Winter Soldier released in theatres and doing gangbuster numbers, we are finally free to shed light on the work of those unsung heroes of visual development - the previs artists. While previs has been around for 15 years or more, its role and importance in modern filmmaking continues to evolve and grow. More films are using more previs than ever before.
In their largest feature film project to date, LA headquartered Proof Inc. spent 14 months prevising the second Captain America film and another eight months handling postvis. They were involved in every major action and visual effects-driven scene including the major third-act finale. They designed the superhero Falcon’s flying sequences as well as scenes with Captain America facing down and destroying a Quinjet, helicarrier launches, helicarrier shootouts and their fall into buildings.
I recently had a chance to speak with Proof’s previs supervisor Monty Granito, who led the previs effort on the film.
Dan Sarto: The new film looks great. Other than a few “Oh, come on now!” stunts, the action looks quite realistic.
Monty Granito: The first words from the directors [brothers Anthony and Joe Russo], the vfx supervisor [Dan Deleeuw] and producer Nate Moore were they wanted the film to feel like a 70s thriller grounded in reality.
For Falcon, the producers at Marvel and VFX supervisors told us for most of the stuff he does [laughs] he has to adhere to the laws of physics. He’s not Iron Man. He can’t just fly around at will in every direction with all these engines. He has to use wing power, buffeting the air with some propulsion. There were points where he was getting shot at by these big ships, getting knocked by big blasts. We couldn’t have him just knocked to the side because he’s just a human being and the blast would concuss him. Just the lip of that blast would microwave him. So, we had to be very careful to show that he was being agile enough to get around them.
DS: Of course. Big blasts would fry him.
MG: But, realistic action was paramount. That was everybody’s goal. That’s what we try to do at Proof. We like to make sure that what we create is not some crazy sequence you can’t shoot. We want to create something that is dynamic…and shootable.
DS: You guys were on this project from the very beginning, handling both previs and postvis. Can you tell me a bit about the scope of your work on the film?
MG: The previs team started off at four and grew to six for a long time before going up to 15. But for most of the previs we had a team of six. It was great getting an early start, to work out some ideas early on. We ended up prevising every major effects-driven piece. In the movie where Cap [Captain America] fights in the elevator, because it’s really stunt-based, we really didn’t previs that. But when he jumped out of the elevator into the atrium and then road the motorcycle and faced off with the Quinjet, we prevised all that. So starting from the beginning of the movie, we prevised the Quinjet shots, everything Quinjet-related, then the boat reveal and then everything else that wasn’t hand-to-hand combat stuff. We kind of left that alone until postvis
The next thing we prevised would be when Cap escaped from Triskelion. That is when he jumps out of the elevator and faces off against the Quinjet. Then after that the next previs is when Falcon reveals his wings. It’s called the Skyscraper Chat. This is the skyscraper shot where they blackmail Agent Sitwell on the rooftop and Falcon grabs him as he falls.
After that we did the whole sequence called The Causeway, where the Winter Soldier rips Sitwell out and there is this whole protracted battle leading to Cap’s reveal of Bucky.
The next previs in the time line of the movie would be the finale, which consists of the helicarriers taking off, all the Falcon stuff, when Cap jumps off the helicarrier edge and Falcon picks him up. So, anything for the finale that has anything to do with a helicarrier or Falcon, we prevised it.
DS: Which is pretty much everything [laughs].
MG: [Laughs] Yes, which is pretty much everything. We prevised the very emotional ending with Cap and Bucky, which was great. A lot of times you don’t get to previs an emotional core of the movie. It was awesome. Then we prevised the forty first floor when the helicarrier hits the building and Falcon jumps out. Cap and Bucky finished their battle. Bucky pounding Cap. Then the engine falls through, knocks Cap out and then Bucky saves Cap. So, all that stuff. There’s a lot of previs in that battle.
It was very cool. We were involved in the story, brainstorming with the writers and the director, editing boards, animatics and previs together with music. It’s the first time I’ve seen this happen on such a scale.
DS: And at what point in the production was this all coming together?
MG: This was all happening before shooting began. Dan Deleeuw, the vfx supervisor, wanted to make sure that the third act was solid before they started shooting. They’ve [the producers] been doing this a while. They know when you have a good third act, the movie is successful. So even before production was started, we had the whole third act in this huge 25 minute edit of boards, previs and animatics. They could tell it was solid before they even started shooting.
The third act was tough. You have these three big ships. Marvel was very cognizant of the fact that Captain America had to destroy these big ships without causalities on a grand scale. Because he is Captain America, he’s not going to level a metropolis. Marvel is great with their characters. Cap would not take down these helicarriers in a way that would hurt innocent people. He wouldn’t do it. So, we had a very specific pill box to play-in. Our challenge was to really make it seem dramatic and look dramatic, so you could immerse yourself in it. It was a big challenge.
DS: All this work you’re describing is leading up to the start of production. Moving into the shoot, how much were you involved onset, during production, leading up to the start of post?
MG: Because this is a CG-driven movie, we were also heavy into techvis. So when they went on to the set, they had the information they needed. I didn’t have to be there. Because so much of the film is CG, a lot of the stuff we did, especially the helicarriers stuff, we just sent the Maya scenes over to the vfx studios. A lot of the Falcon stuff, a lot of the helicarrier stuff, even some of the Cap stuff was never going to be shot.
DS: Did you have people doing work at ILM and Scanline?
MG: No. We would just send them files. What we would do is, we would take our Maya scene, export everything, send it to them and they’d bring in the Maya scene and do their conversion. Then they’d have the camera and the ships and then they could make everything look great. A lot of their work matches up one-for-one with our previs. They slowed some stuff down for size and gravitas and made it look better. The previs is so low res that if you moved it that slowly, it would look boring. But when the ILM people get a hold of it and they put all that detail in, then you can really let it move like the mass would allow it to move.
But in terms of the composition of the action and the choreography, the stuff that we sent to ILM is one-for-one. Every Falcon shot is pretty much exactly the same. The same thing with the helicarrier shots, which is very gratifying. I remember when we were getting helicarrier shots back in preparation for Comic-Con, because the cameras have been so locked down in previs by the directors, all ILM’s revisions were about the look. It made things a lot more efficient that they didn’t have to go searching for cameras.
DS: Based on my understanding from various projects, there is often a lot of movement and change from what previs provides and what the VFX artists ultimately create. It sounds like that was not the case here from a choreography standpoint.
MG: We wanted to be with the VFX supervisor and directors as much as possible. If we could see them every day? Then every day. Twice-a-day? Then twice-a-day. We wanted to help them make their movie. We wanted to give them ideas that would help them make their movie. That is why the previs came out so exact, because they made “their” movie in previs. When the natural thing happens and the camera start to change they said, “Wait, no. Go back to the previs. We have the previs, it’s timed right, lets go back to that.” They were already happy with the previs when it came time for shooting and then post.
DS: So setting aside postvis for a moment, it sounds like previs and techvis were used extensively across many areas of the film, to really lock down the movie wherever possible. I imagine having good techvis probably saved the producers a fair amount of money and time regarding what they actually needed to get on camera.
MG: Yes. Here’s an example. There is this moment when the Falcon is running from the helicarrier as it slams into the building. Because we had given them techvis data in terms of the speed of the helicarrier and how fast Falcon was running, how much the ship was gaining on him and how much area it was covering, they could actually rig the entire thing for real. If you don’t have that data and if you’re just kind of guessing, then you can’t rig up the shoot for real. You have to do it in post. It’s too wasteful to keep rigging it up, starting it, rigging it up, starting it. They had all that data. They had the shot planned out to a tee. We even gave them cameras that weren’t in the previs to show them what you could get if you just had a camera man standing there. So, they rigged the entire scene up for real and actually destroyed the whole set. They had one chance. But because they had a lot of planning for the set already mapped out, they were able to take that chance.
DS: How much previs was still left to do once the film actually started shooting? At what point does previs finish and postvis takeover?
MG: That’s the funny thing and why times are changing. There never is that point essentially. When I left the project well into postvis, Ranco Todic, one of the main guys from the team, was still prevising shots. Throughout most of the postvis process, we were always still prevising certain shots. You know, we’re going to shoot this tomorrow, let’s try this other angle and see if we can get something better…They’re trying to be as efficient as possible, but they also try and get as much work in as possible. I don’t think they ever cut it off essentially until the film was shot and in the can.
Even then, you can always make something better. They didn’t care about cost or how long it took or how many versions were done. They just wanted it to be better. Let’s find the one good shot. With Dan leading things, we’d do an entire sequence and he would say, “Keep that shot and do all the others over again. That’s the best shot for now.” Well, let me do it over again. Now this shot is the best shot. We’ll keep that one. And so it was just doing that over and over again until you have a sequence just as good as it can possibly be. It was great being part of something where all they cared about was making everything better.
DS: What were some of the main challenges you faced on this show?
MG: You may laugh, but I didn’t think of it as challenging. We felt like we were part of the family. Everyone on the movie, we all felt like we were making that movie with our bare hands. As previs people, as freelancers, we’re used to coming onto and off of movies. But Jen Underdahl, the Marvel producer, would tell us, “No, you are part of the family, you are part of us.”
So the only thing that was really challenging was when we were slowing down and having to let people go, or if we were ramping up and needed to bring on five artists in two days. It’s really hard to find five good people in two days.
We did shield replacements. You may not know this but almost every shield, even the one he’s wearing on his back, are CG. It’s crazy. In postvis we were doing a lot of shield animation and set replacement in the dome. The Falcon stuff was the biggest creative challenge because you know no one had really fleshed out a character like that before in live action. Dan pointed us to some squirrel suit guys as reference. He wanted it to be shot like that in terms of the visceral nature of the action.
Funny story. The entire Falcon sequence where he flies down to the deck, shoots the Quinjet, chases and jumps off the edge and goes into the ball, that whole thing to some extent didn’t really exist. I gave a test to two of my guys, George Antzoulides and Eric Benedict. How does Falcon fly? Guys, go make Falcon fly! Do whatever [laughs]. But, we were told, he can’t flap. Don’t ever make him flap. We had three concept drawings. Eric, you do this concept drawing, George, you this concept drawing and make a sequence. So we just started cutting something together and it began feeling very organic. Dan was very supportive. So we did this big test. It’s just a test for Marvel and for Comic-Con. How does Falcon fly? And they liked it so much that they made it into the sequence.
Because the work was so collaborative, we’d see them every day. They could give us input all the time. We’ve worked on other shows where we had to work alone. And even on this show sometimes we had to work alone. So if someone puts you in a corner and says, “Make me a cool sequence,” you make them a cool sequence. But you’re not going to take the same leaps as you would if you got input and approvals right on the spot.
DS: That’s a completely different creative dynamic.
MG: Yeah. When we began the Falcon sequence for the finale everyone had creative input because we were all in the room together. Dan wanted Falcon to have "star destroyer" shots. I liked panning shots while the Quinjet was chasing him. Eric wanted to have Falcon kick a guy and then get surprised by a jet. George came up with the closing the wings moment. We were all together collaborating on a daily basis and the best idea won.
DS: For the most part you’re working with low res assets, turning around stuff very quickly. What types of tools are you using? How are you presenting materials for the director, the producer and the vfx supervisor to review and interact with?
MG: There’s this great tech director at the office by the name of Anna Lee, who builds the tools we use most often. She built this kind of render tool. So essentially all we have to do is worry about making the cameras. In the past to make a complete shot you would have to render it, then place all the data into the camera and then re-render that out. So you’d take three or four steps to make your finished movie that you would then put in an edit. What Anna built is this tool where you make as many cameras as you want, you press a button and it spits out a movie for you. Properly named, properly formatted with reticle, with lens information and with focal information. She spent a long time developing this perfect reticle that not only can simulate any live action camera, it can also give you any piece of data you or the production wants on the reticle in terms of speed, focus distance, focal length, plus the height of the camera. So all the artist has to do is make the camera, press render, go get a coffee and come back to a bunch of movies that can be delivered.
She also designed this proprietary toon shader that we used on Hunger Games. It’s beautiful. It’s a real-time Maya toon shader. No rendering. It’s gorgeous. It looks fantastic. When we went onto Captain America, Marvel asked us to do black and white previs because it fits in with the boards and animatics pretty well. So all the Captain America previs is in black and white with a couple of hints of color. That worked well for a couple reasons. First, it made things super quick because it rendered real-time and was superfast. It also allowed us not to worry about things that were not our job. The color scheme of the film is not our job. Sometimes you have to put in colors and then there’s a kind of temp love and then the art department can bristle a bit at us making color choices. So we didn’t have to think about what color to make the street, what color to make the car…make it grey, make that car darker grey, make that color a light grey. We should only be concentrating on what we actually deliver to the final product of the movie, which is composition and storytelling.
DS: On the other hand, does providing low res visual materials without lighting or color hinder or otherwise adversely impact the ability of the directors to assess how the story is coming along?
MG: The animation, storytelling and composition should be enough to tell the story on its own. High res visual materials, color and lighting are great, but we feel like if a sequence works in low resolution made as efficiently as possible, it will really work once it's a final movie. If your storytelling is good, if your animation is dynamic and if your edit has texture and momentum, you can immerse and inspire the filmmakers without a lot of bells and whistles.
DS: Are you working primarily with assets provided by the art department? Or, are you creating all your own assets from scratch?
MG: We get tons of stuff from the art department. All kinds of stuff. It used to be that we created everything, but more and more the art departments are filled with 3D designers now. So, we get a bunch of models from the art department. We just have to re-texture them, optimize them, put the shaders on them and get them into our scene files.
DS: From the postvis standpoint, are editors dropping finished shots into the previs? How is your previs integrated into the final edit? How does the vfx work and editing come together through postvis? Are you going back and forth with the editor at this point? Is the idea that you are re-editing previs with live action materials so that the editor can start assembling a final cut before actual vfx work comes in? What is the goal of the postvis?
MG: When the editor comes on, we essentially stop editing. So the main point of the postvis we did was first, to see if the main story beats worked. See if they got what they wanted from what they were shooting. Once you put a plane in there and a building, you could see if you got what you needed or if you needed to reshoot. See if the story played out. Then we add the effects based on what the editor calls out. The editor says, “I want a plume of smoke there,” or “I want the sparking here” or “I want the plane to crash now because I have a sound hit that is coming after that.” So, essentially the editor becomes the postvis director.
In previs, the main people we work with are the directors and the VFX supervisor. In postvis, we are essentially only seeing the editor at this point. So, we completely change gears. Our entire purpose is to make his edit look as good as possible for screenings, for the directors and for the producers.
Dan Sarto is Publisher and Editor-in-Chief of Animation World Network.