Believe it or not, the technology does not yet exist to make 30-foot tall transforming battle robots, not in a practical manner anyway.  So the big bots of Michael Bay's Transformers: Age of Extinction are left up to ILM's visual effects supervisor Scott Farrar and his team of up to 350 artists.  I got a chance to talk to Farrar, along with a small group of journalists, while on set in Detroit, which was masquerading as Hong Kong before Bay blew the whole thing up.

In addition to the movie's robot redesigns, Farrar talked about the difficulties of shooting in multiple formats, advantages of shooting practical effects on set and digitally augmenting them later, and the process of digitally recreating buildings and entire cities.  Hit the jump for more from Farrar on the behind-the-scenes process of visual effects.

Question: Lorenzo was talking about the speed at which you make these films. How fast is it for you? How does it change each time? Are you in post as it’s still in production?

Scott Farrar: Yes. Every one seems to be faster. There’s less and less time for the number of shots. It’s all about how many shots you do in the movie. In this one, we have more shots to do and the complicating factor for a lot of the work is multiple formats. We’re shooting in IMAX, IMAX 3D … I’ve shot IMAX film, I did a bunch of the aerials and location stuff for one of the sequences in Iceland; that’s all on film. Then I shoot a lot on digital as you know. And we’ve got anamorphic, we’ve got spherical, we’ve got all these different formats that have to be intermingled. So we’ve got single lenses that have to be converted to stereo, right? And then we have stereo cameras also. With all that, it’s a big mess to work it out, just to manage the data, right? And then doing the shots, we start in very early. We started building new models of robots, because every time we get new models … this one has something like 18 new robots, 18 new characters. It takes us at ILM generally speaking about 15 weeks to build the geometry, the shapes and the whole thing, of a model. Then another 15 to 20 weeks to build the rigging, which is the skeleton that holds all the pieces. It takes a long time because it’s not as simple as an arm, it’s got all these pieces attached that, later on, have to be in the hands of the animators, they have to be able to control all the things that spin and all that cool stuff. So when an elbow is close to camera, you see all those things turning. Then, doing paint, textures, and all that sort of thing, long before we even start doing the movie, otherwise we can’t get all the robots done in time to actually put them into production and start the animation.

How much is that total per robot?

Farrar: How many weeks? I guess the real answer is that we still alter them after they come into production. I don’t have a fixed number, but this is a creative task after all. You start with flat artwork, and once you turn it into three-dimensional shapes in the movie, you always change things because the flat artwork might look really cool from the front and the back, but when you look at it from the side it doesn’t look right. So you play with the proportions. Then you put it into a shot and then you see it with a different lens. The lens, just like humans, a wide distorts and a longer lens makes you look better, well with some robots they look dopey when you put a certain lens on. So then you’ve got to change the configuration.

transformers-5-michael-bay
Image via Paramount Pictures

Guillermo del Toro did a presentation after Pacific Rim where he talked about streamlining the pipeline by making sure at the very beginning of the process it was exactly what he wanted. Michael seems like a guy who has a lot of certitude. Is there a lot of revision in this process, or is it well defined as soon as you guys are starting?

Farrar: Well, Michael’s really great on designing things that look really good and then he doesn’t change too much. He’s great at deciding on the look. In essence, a great cinematographer, in a way. He knows how to do good shot design, how to make a cool move, and how to make a cool shot. Generally, once we come up with the design, we leave it. We shot one of our first sequences in Texas and we’ve already got animation in play right now, and we’re seeing Optimus in a bunch of the shots and he’s got new designs on him. Each and every character has updates from the previous films. Bumblebee, Optimus, all of them. It’s just now coming on. We’re already doing shots, so when I’m on set, I’m getting shots fed to me all the time from ILM that the layout’s been done, the animation’s been roughed in. We’ve still got a few shots where we get some rough lighting on them and you say, “Eh it doesn’t look right yet.” Once the character gets into the movie, then that’s when the work really begins.

How many people do you have working on this film?

Farrar: We have some core people, but I think there are only about 50 or 80 people right now. By the end of the movie we’ll have about 350. We also have our Singapore office. We’ll have a pod of people working in Vancouver, and we’re working with people just outside of Bejing, so it’ll probably be more than that number. This crew is over 200 people. You don’t realize how big our thing gets later on.

Can you talk a bit about the animation process for the robots? What are some of the challenges?

Farrar: As you may know, just for acting, things like that, generally speaking it’s key-frame animation. We do do mo-cap sometimes when we’re trying to do new types of stunts. Lots of times we’ll do studies of the stunt men. Stunt men will come up with new ideas for a stunt sequence, and we’ll videotape that, or record that, and then we might recreate it back at our place with mo-cap suits. But most of the time it’s all key frame, that’s what we end up with because the best acting comes out of the finesse that the artists of Scott Benza and his crew, he’s the animation supervisor, they’ve gotten just fabulous at doing this stuff. You can imagine, we’ve had eight years of training doing Transformers, so everybody’s gotten better and better, and we have a lot of people that are on staff that work on these films. They love these films because every shot is fun to do.

transformers-4-age-of-extinction-michael-bay-nicola-peltz-set-photo

There have been some really cool blended shots between the CG and the practical. One that sticks out is Bumblebee’s transformation in the middle of driving away while Shia is ejected out. Anything crazy that you can talk about as far as blending the practical and digital?

Farrar: I call those switcheroos where you might real/CG/real/CG. We have several series of those where it goes in and out a couple of times, but it’s still really, really difficult to make a human look really good. Whenever we can, we’ll probably do a lot of stuff with the actors where they’ll get grabbed, or cradled, or held by one of the robots … I’ll try as much as I can to get that against a green screen or something because it’s a lot of work to make … even like on the last picture, we had a 100% CG version of him for some close-up things, and it’s just really, really difficult. Just the slightest shading on the nose will make him look wrong, or where you don’t have enough ash or debris or whatever was in the scene. Just colorations. It’s so bizarrely difficult. At a distance, it’s easier, and we’ll have those. Cutting-edge though, it’s a learning process; we’re still learning.

Do you have any bigger or more elaborate things that you’re doing with the Transformers themselves?

Farrar: There are some things I can’t talk about, let me think of what I can talk about. [laughs] I’ll tell you what I can talk about. It’s not always just the robots. This is kind of fun stuff where you are right now, but this is supposed to look like Hong Kong. A lot of the fun things that we’ve learned from the past movies that you may not know are recreations of cities where we do a lot of virtual photography, still photography, to recreate all the buildings, like the titled building in the last one. That was put into, actually, three different locations in Chicago. It wasn’t always the same. Every time you looked at it from a different angle, for a variety of reasons, we had to change the location. Those are all virtual recreations of buildings. What’s fun is not just robots, but also the set extensions. Now we’ll photograph this stuff in here and some of these buildings will have to be replaced, or different backgrounds. So we have a whole team of people who spend a few weeks in Hong Kong and shoot every side of a building top to bottom so that we can take those as pieces of geometry and put them into the background. I think that’s fun.

What do you like about coming back to Michael and the franchise? What excites you?

Farrar: Of all the movies I’ve worked on, this is one of the best teams and relationships between ourselves, ILM, and the studio, and the producers, and the director, and all these people who have been working now for years, and it’s like any team that relies on one another. It’s pretty fabulous. Some of it’s non-verbal communication, everybody kind of knows what to do, but everybody’s truly taking risks all the time. Let’s try something new. Let’s come up with something we haven’t tried before, we haven’t seen before, or we haven’t done before. This group is dedicated to that. That makes it fun.

michael-bay-transformers-6
Image via Paramount

Obviously Bay is known for spectacle, these massive explosions… The robots aside, what’s that process like for you when you show him a shot? What’s his feedback on it? “Bigger! More fire!” or what’s his take?

Farrar: A new shot or a new idea is like Christmas every day. The problem for Michael will be, he’ll turn over a shot and say, “Get started on this shot,” and then he won’t say anything for a few weeks, and then, “What’s happening with that?” and I say, “It takes a while.” I’m the same way. I can’t wait to see what kind of things we’re going to do. Going back to what’s new and different, that’s not just characters. This movie, we’re shooting green screen stages, you know all about extractions and green screens and all that. We’re shooting this with smoke … we’re doing everything wrong, everything, and we’re just going to make it work because I keep finding that the dirtier it is, the more messed up it is, the better, because if I try to add that later, it never looks as good. So we are constantly breaking the rules. It’s sort of a general theme. It’s really not so specific, but in terms of the evolution of visual effects, that’s the only way we can make better-looking shots. That’s what I’m trying to do.

So you prefer to have a practical foundation that you can enhance and build off of digitally?

Farrar: Absolutely, absolutely. We were just talking about some of the things we have to do here today. I struggle because I can’t do miniatures the way I used to because, why? The camera’s always moving. Remember on Transformers where the robots flew through the building? That was a miniature we shot. To do that, we were doing about 120fps, frames per second. That meant that the camera on the rig had to travel 50mph within the stage, which means it had to hit speed within 12 feet and ramp out in 15 feet, all driven by a cable run by an automobile and an axle. All these contraptions that you get into that there’s a point that physics prevents you from being able to do that type of shot. So with all these moving cameras, these cameras are moving all the time. You see it, right? Cranes everywhere. I can’t shoot them in there. I can shoot elements and put them into the miniature, into the scene, but now, that’s why we turn to computer graphics work so much. It’s got to look good, it’s got to look real. Or we shoot as much as we can real here, and then bits and pieces will be plugged in later.

transformers-5-michael-bay
Image via Paramount

You were talking about all the different formats, with different resolutions, how does that work when rendering these Transformers and effects work?

Farrar: We try and stay at 2K, but we learned a lot about that and what to do when we made the second one, Revenge of the Fallen, because we shot certain sequences 100% on IMAX film. It wasn’t 3D, but single lens, large format, but as I recall it’s eight times more information that you’re dealing with as opposed to a 35mm frame. It’s a huge, logarithmic jump in the information you’ve got to carry. And so it goes back to the render farm, and the capacity of the render farm … everything gets cranked up and some of those frames took 36 hours a frame on the highest speed render at the time, so the answer is, technically, it always creates new problems that somehow or another you have to solve. Now we’ve got IMAX stereo, so whatever I’ve just told you is now doubled. That goes on and on. It’s crazy! [laughs] It’s insane actually to do this.

transformers-age-of-extinction-poster-dinobots