‘Mitchells’ animation supe Alan Hawkins discusses the insane robot rigs on the film.
Among the many animation challenges for Sony Pictures Imageworks on The Mitchells vs. the Machines, one was the stealth robots seen towards the end of the film. These moved in unusual ways, and at times—spoiler alert—get sliced up. For that, the VFX and animation studio developed new tools to make that event happen.
The film’s animation supervisor Alan Hawkins discusses the approach here, along with the other key animation challenges along the way.
b&a: In the film there’s this cartoony animation style, but also very realistic. From an animation point of view, what did that mean for you going in?
Alan Hawkins: Yeah, that’s a good observation. The proportions of the characters faces are obviously very cartoony and even then just their overall body shapes. But the acting choices, mechanics, limitations, physicality, that kind of thing are much more grounded.
That was something that Mike Rianda wanted from the very beginning, early conversations set the tone very broadly for that kind of thing. He was very specific about wanting to fill it with as much rich, observed details as possible. Whether it’s how someone engages with the environment they’re in like that scene with the chair or how you sit in a chair or playing with a salt shaker at the dinner table while you speak. Also, how people tend to have conversations while doing other things—sometimes you’re unpacking groceries, whatever it is. You’re not just standing or sitting at a table facing each other and speaking all the time.
We use a lot of video reference and then chose where to exaggerate things. Almost never did we do less than was in the reference. The reference would usually be a very realistic take on something and human subconscious gives you a lot of surprising answers sometimes that you wouldn’t have thought of on your own, if you didn’t just do it spontaneously.
Then we looked at ways to push something like the way a character does a little smirk or a weird little half frown thing. Then we would amp up those key moments that were bringing something special to it and sometimes there is broad cartoony moments, but as a general thing, it was about trying to make it seem familiar and mature in a way, and to really shy away from common animation bag of tricks. So it felt more like a living, breathing characters, people.
b&a: Early on, were there discussions with Mike or others about whether generally this should be something animated on ones or twos or otherwise?
Alan Hawkins: Obviously following the heels of Spider-Verse, Mike was very excited by all the things they were doing and said, I want all of that, I want to make it special like that. We did have that discussion. It didn’t seem appropriate for this film to do anything other than fully fleshed out. It really worked for Spider-Verse. It helped reinforce the comic book aesthetic, but at the end of the day, we couldn’t really justify any reason to not do it. We felt like the art was pushed enough in its own way so it wasn’t really necessary. Even though I do love seeing it in other projects, it just didn’t have a place for this one.
b&a: One thing I feel, but I don’t know if this is true, that sometimes because of the zaniness of the film, characters and their models can just do extreme things. Were you were taking the models and doing extra things with them at all? Were they rigged in any special way?
Alan Hawkins: I think that feeling probably comes more from the circumstances more than anything. There were some things that we did, some unusual, I guess you could call it groundbreaking stuff we hadn’t done before. But not with the human characters really, they were pretty standard. As far as the technology goes, we had a handful of floating ink lines on the faces that we inherited from Spider-Verse.
They used it much more on that one, but ours was a much subtler, little nuance. A lot of the ink line and render effects were done in post by the lighting and comm teams. So animation only handled a few key ones on the faces, but no, if you’re getting that feeling, it probably has to do with just the situation.
b&a: What was that thing you mentioned about the non-human characters, where you were doing things that you hadn’t done before?
Alan Hawkins: This was for the stealth bots. The way we animated those was completely new. When we first saw the designs, we knew we wanted to do something non-bipedal with them and didn’t want to just see them walking around or anything. We started exploring different ways of having them warp around and break apart with organic-looking stuff and geometric-looking stuff. We liked the geometric stuff.
When we started, we developed these two tools to basically break them apart differently every time. The rig for those guys is extremely basic; there’s a single joint because they’re rigid, there’s only a few shapes to them. It’s about as simple a body rig as it can be. But these two tools allowed every animator to dissect them differently every single time. There was no ability to share poses. There was no undoing once you started to cut a character apart, it’s a very strange process and very unstable too.
A lot of very smart people in the pipe and backend departments made it so that this very unstable geometry would render. Because it was, it was very glitchy. The tools would cause it not to get too technical, but every time you would move it, it would create a whole new set of verts. There was never the same number of verts from frame to frame. And it would basically just like glitch out, disappear, invert at random. And so the animators had to work around them and sometimes getting notes on a shot meant redoing it from scratch, but the tools were very fast and people got very used to them that a redo on one of those shots might only take a day or so.
Those were really cool and exciting, and it was largely driven by the Imageworks side. A lot of the time things are delivered to us already thought out and very meticulously planned from S.P.A or any other client. For this, we jumped on it real early when we saw the art and Mike and Lindsey allowed us to, they gave us a chance to make our pitch and it all worked really well and they, they love the results. That was a pretty rare opportunity. That was really fun and rewarding because it was a whole strange set of tools and a whole weird way of moving them around outside of the standard rig situation.
One of the two main tools we developed for this let you draw a line through the character and it was just a straight line. What it would do is, through the camera view, slice the geometry in half, cap off the top and bottom part, apply the correct naming so that way it could travel down the path. It was just like a series of 10 commands that every time you clicked and drag and cross, it would make you this little separate piece that then could be animated. Then also being able to animate that piece, cut it again, and continue to animate the two smaller pieces. It was this kind of nesting dolls pattern of a geometry hierarchy. That was the easier of the two, because that was more like just a macro of commands.
The other one was using booleans, which is the much more unstable part. The animators would create either a subtractive shape out of any geometry they want or an additive shape. So you could basically cookie cut anything out of something like a 3D cookie cutter and translate that away, but also scale through the volume. You could activate one of these shapes and let’s say it’s the soccer ball in the chest or something, and then scale that, you create a little cutout in there and you can move it through the head, through the legs. It was very weird spatially, but it gave us really cool effects.
b&a: When Linda Mitchell goes ferocious on those robots, I wanted to ask you, as animation supervisor, how you would break that down for your team to create that craziness?
Alan Hawkins: This movie had very particular storyboards. Mike was often quite married to them. There was always room for pitches or new ideas, but the default choice is to look at the boards and see what he’s done, because he loved his storyboard artists, Guillermo Martinez and Hanna Cho and lots of other artists. So a very normal part of the rounds or daily conversation would be looking at the shot and then saying, can we see the boards? Can we check the board? It’s always about checking back to see what they were because he worked on a movie for years before it got to us.
There were certain things that they’re just very used to and certain really important moments like that were largely figured out before they came to us to a certain degree. We were always adding and plus’ing things where there was room for it. But I would give that credit to the story artists, there was a lot of thought that went into those kinds of moments. There’s lots of places where animators brought their own special version of an acting choice or things like that, which Mike was extremely open to.
b&a: I asked Mike Lasker about Furbies and I felt like I should ask you as well, Alan. What particular challenges in animation did the Furbies bring?
Alan Hawkins: It was important to capture the way their little motorized bodies would work, especially when we’re doing the very Furby-esque motion when they open their eyes for the first time when they were introduced and the weird little talking patterns. They’re not very elegant, they’re kind of jerky and layered. That just came from observation. We would look at them and see what they would do and then recreate it because it’s only the iconic thing that it is when you make it work exactly the way it did. It’s a very recognizable pattern of motion.
Now, the giant one is obviously different. It doesn’t have that degree of matching the observational stuff, because it’s much bigger and faster. But it’s just having fun and working with the limitations of what that thing is. They don’t have individual legs, they just kind of sit on a disc and there’s one little motor that moves it forward. So we had to invent the idea that it could kind of lumber from side to side, but that’s the kind of thing that you want to see, something that size do as it moves. That’s what you expect it to be. Then from there is just embracing the absurd shape of the thing and letting the limitation almost be the inspiration or working within that.
b&a: Did you also have to deal with any limitations for the Pal robots, the white robots, because they’re joint based?
Alan Hawkins: It’s always nice to have a character like that. It’s a very lightweight rig, obviously they’re rigid and don’t have a whole lot of extras to them. So working with those characters is a nice balance between the very complex things and the very simple things, which is what they were. Doing scenes with them was always nice because they were very fast. We played with the fact that they’re rigid and you could spin things indefinitely and play around with that kind of glitching and stuff, which we did. We did a lot of explorations about how they move from the very beginning, like having one turnaround without actually turning around like individual parts spinning to kind of reverse and face the other direction and stuff like that. Which I think a couple things like that made it into the movie.
Then for the glitching ones, for Eric and Deborahbot, that was almost like another layer upon layer upon layer thing, because they’re robots, they’re glitching out. They also have a personality. There’s also jokes. There’s all kinds of stuff at play in there that had to be just right. So it was finding this weird balance, and it’s almost always, the challenge is always comedy. The jokes are the hardest stuff, mechanics doing a cool landing or a roll, that’s not as hard, but getting a joke to really move that’s where it gets really hard. That’s for every character almost all the time.
Comedy is sometimes a lot harder than action, but it was always really fun to play with them because it’s all pantomime, there is no expressions, so whatever our hand gesture or body pose or whatever it is, has to tell that story. So just a whole other flavour to deal with.