Create: The workflow seems overwhelming. How do you manage the output of so many different teams?
White: The way the work flows through visual effects production, it’s quite a complicated machine. There’s a lot of interdependencies between the various departments. It’s like home construction in a way. There’s a department that pours the foundation, a department that sets the framing, there’s one that does the electrical and the plumbing. And into the mix you have creative feedback from the client—the director—that you’re constantly responding to or adjusting to.
We often start with a plate that has actors and a set that we’ll be working with. So the first step is to lay that out and do our camera tracking in our layout department. All of that data is then passed along to animation, and animation starts placing all the characters into the scene and animating them.
They get all the animation input from the modelers, who essentially create the character—and not just the static character, but they have a whole library of, say, facial animation, and the modelers are responsible for creating all of those facial shapes. Especially now that we’re doing a significant amount of not just hand animation, but actually putting helmet cams on the actors and capturing those performances, that process involves building an entire library of the actor’s face, of all the different poses that it can hit, so that when we go to record the footage, that’s all transposed onto the model. That could either be onto a digital version of that actor—like if we’re doing a stunt head replacement—or transposed onto a creature, in which case that actor’s performance is then made to work on sometimes a very different facial anatomy. But the most important thing through all of the animation is really capturing the performance that we’re looking for.
Create: In the opening sequence of Terminator: Dark Fate, you had to create young versions of Sarah Connor and Arnold Schwarzenegger. How did you do that?
White: Recreating them was incredibly difficult. Human likenesses are one of the hardest things you can do with computer graphics, and especially these characters that are so well-known and well-loved from Terminator 2.
We tried to keep the process at first as scientific as possible, in terms of doing scans of real skin and measuring blood flow and looking at how pore sizes change as the face stretches and compresses, and then spending an incredible amount of time just studying reference images, studying over the actors’ faces, at that age and at that time, and making sure we could accurately recreate them.
We had a really fun moment where we showed a work in progress to Arnold and he said it looked pretty good from the nose up, but the mouth shape looked wrong, it was too puckered. It was interesting, because that mouth shape had been modeled from a reference we had of him, but in that particular camera angle, he was right, it didn’t look right.
It just shows you how tricky the human face is, because it is so lighting- and camera-dependent. We actually went and found another pose from one of the posters of Terminator 2, and modeled all the detail into the mouth shape that we saw in there, and that ended up being what was in the film. So it’s very challenging, even when you have a pretty good-looking static model, to be able to recreate the likeness of an actor, once the faces are moving.