THE CUTTING-EDGE TECHNOLOGY OF ALITA

THE CUTTING-EDGE TECHNOLOGY OF ALITA

Alita: Battle Angel is shot natively in 3D, rather than shot in 2D and converted to 3D in post. "Because Lightstorm was at the cutting edge of the 3D renaissance with the first Avatar film and part of blowing it up to what it is right now, we have a reputation that we have to maintain of doing the best 3D," explains James Cameron. "To me, the best 3D is native 3D. [Shooting 2D and] converting it later is automatically giving yourself a maximum of a B grade. The only way to give yourself the possibility of an A+ grade for 3D is to shoot in 3D. Not that many people do it, but the directors that do it — Martin Scorsese and Ridley Scott and Ang Lee — are amongst the best filmmakers. That should tell you everything you need to know about whether you should shoot native or not.
	"Robert Rodriguez had shot native 3D before," Cameron continues, "so he was completely familiar with the rigs that Vince Pace and I developed. He already knew how to do it. That was almost a non-conversation: 'You doing native 3D?' 'Yeah, of course.' Which would have been a multi-week conversation with another filmmaker!"
	To understand how 3D works, "Think of 3D as being a pair of eyes," animation supervisor Richie Baneham suggests, "but each eye is a camera. So you're rendering the subject [being shot] from two discrete perspectives, and the difference between those perspectives depends on how far apart the cameras are placed (the interocular distance between the two). Each camera is viewing the subject from a slightly different angle, with the subject of their perspectives being at the convergence point. the %%dy the human eye works, and therefore how stereo [3D imagery] works."
	Rodriguez has decades of experience with both CGI and the native 3D process. From camera operating to editing, he has done virtually everything involved in filmmaking at one time or another. "I'm used to doing all those jobs," Rodriguez acknowledges, "so it was great having Weta there on set [for Alita]. When I did Sin City, I was the Weta team. I had to come up with everything, because it was a very scrappy production, where you're the whole crew So having these stellar resources was just unbelievable, and allowed us to move so fast. I'd say, 'Guys, you want me to get a clean plate of that, because I would be getting it to help myself later in post?' And the Weta guys would say, 'That would be great. But we don't want to bother you by making you do this.' 'No no, I know that gig.'
	"We had to rig cameras everywhere for performance capture," Rodriguez adds about the CGI element of the shoot. "You have to pre-rig everything, to be there getting all the data, keeping track of each shot, because every shot is an effects shot. It was just so smooth. I thought for sure that whole system was going to slow us down, because it was such an army of people, but Weta have just got it down. We came in on schedule."
	For Alita, Rodriguez explains, "We shot with the Alexa 4K dual-camera system — the best we have — for true 3D capture. You've got the A-camera on a crane and another team on the second camera, usually on a dolly or a smaller crane. Then the 3D guys are a different department. They're in their own tent, adjusting and doing the 3D as its shot. With the DP [director of photography] over in his tent, talking to all of them, telling them what to do. It was different from when it's just me running and gunning with a camera. It's a lot to organize. I'm watching and responding and giving notes as it happens."
	Although Alita marks the first time Rodriguez and Cameron have collaborated directly. Rodriguez has been using Cameron's 3D techniques for a long time. "I started shooting digital on the Spy Kids movies." he notes. "I was thinking, 'Digital is capable of so much. I wonder if I could just strap two of these cameras close together and shoot on greenscreen, and then in post, slide them closer together and create a sort of 3D?' I was doing a bunch of tests. Then Vince Pace [Avatar director of photography: LA, who also did additional photography for Spy Kids 3] said, 'Jim's already built this system.' I got Vince, who had designed these cameras with Jim, to rig up an over-and-under system, where we could shoot through glass and do the traditional 3D, but with digital. so we could see on set if it was working or not. That was pretty awesome and game-changing."
	Rodriguez is also very happy to be working with director of photography Bill Pope. "I thought the work he did on The Matrix was stellar and visionary," the director enthuses. "I love to shoot with DPs. I learn a lot of good tips, seeing how they solve a problem. Bill was a joy to work with. The hard part when you're shooting a 3D movie and operating [the camera] yourself is that I'd have to run back to the 3D monitors, have them replay the take. watch it in 3D, then go back on set. It slows everything dowl. Bill was just so good that I don't think I ever operated the camera once, I was always watching the 3D."
	There are often choices to be made as to whether something will be shot in practical 3D — using greenscreen or motion-capture reference — or be rendered entirely with CGI [computer generated imagery]. Rodriguez says this is usually "an individual call on each shot. T'here's stuff that we try to do practically so at least we'd have reference. I like to have a reference of what an arm movement would look like, for example. We can always replace it if we have to, but we try to do stuff as real as possible."
	Concept artist Steve Messing says of his part in the development process, "We do all sorts of things. We build 3D models, we have a drawing tablet that we can use to sketch, we use photography we'll shoot photos of textures and warp them into perspective. For Alita, we even built primitive 3D models in the computer of the street set. We would take photos from a certain perspective of little Styrofoam core models of what we wanted to build for the set and paint over them to create an image. The visual effects team then made the decision whether to build models for it. Some stuff, they don't necessarily need full 3D models, it's just used as a frame of reference for the visual effects look."
	The concept art informs the digital design, but it is not literally incorporated into the CGI. "We always build in the computer," Weta visual effects supervisor Eric Saindon notes. "Concept art is primarily [for reference], because it's only looking at the subject from one angle." However, when the subject turns within the computer in three dimensions, the new angles may- present different aspects. "That's part of the process. We always have to build everything low-resolution first, a quick  CGI model, so that we can see it rotating, make sure everyone's still happy.‘‘ h it. usually make some changes, and then get back to making sure all the angles work for everybody."
	Asked to give a layman-friendly explanation of how 3D builds are done within the computer, Weta senior visual effects supervisor Joe Letteri says, "Inside a computer, you start with dragging points and lines around, and you line them up into a 3D model. It's almost like building them out of a wire frame, where you start connecting points, as if you were drawing a triangle on a piece of paper. But then you rotate it around a little and draw another triangle that connects to it, and draw another one that connects to it. In this way, you start building up shapes. You start stitching all these pieces together and building it up in the shape of whatever you want. You have a lot of design tools to simplify that. You can take a circle and extrude it and make a cylinder, you can take a square and extrude it and make a cube. You are building and designing and sculpting inside the computer, so everything has three dimensions."
	"The sketches and the artwork we get are a visual guide to what we can do, for example, telling us how big certain things are," Saindon says. "Then once we actually get it into our world, it needs a lot more figuring out in actual 3D space. We can't just pop random buildings into a city and visually make it work. Because it's a 3D movie, we have to build the entire city to get the visual representation of what [visual consultant] Dylan Cole has conceived in his concept art, and achieve the correct depth and the correct atmosphere, so we can see the city properly. We start with an empty space, then build our pavement, our sidewalks and all the buildings. We build it using different tools we have, to create the types of buildings that Dylan has in his artwork, and that Steve Joyner has built on set, matching the actual set construction.
	"Because we're capturing the Alita character in all the different locations, we had to set up a performance-cap volume in all the locations," Saindon continues. "We would always go in the night before shooting, set up our cap cameras, so that we always had a volume to capture Rosa's motion. A volume is basically a space. We have a hundred or so cap cameras on set, all pointing at the same space, and that space is the volume."
	There are different challenges in creating a human-style character like Alita as opposed to more cybernetic individuals like Grewishka or an environment in CGI. "Grewishka and the Motorhallers all have to move in certain ways. The face — the eyes, the mouth, the nose — is obviously the first place you look. But the physical movement of those characters is how you sell them," Saindon says. "If the design doesn't allow us to move the arms or the legs in a proper way and you can't sell the movement, the viewer will never buy those characters."
	Regardless of how sophisticated the technology becomes, the creative human factor remains vital, Letteri notes. "A lot of it still has to get done by hand, especially for these characters. They're very individualized, and they're highly articulate. So you've got this unique design for an arm that's built out of old parts, but it has to function like a human arm does. It still has to be able to reach and twist and grab, So we have to think about how you're articulating this, as if you were going to build that as a true 31) robot arm in the real world, We just never take it to the point of manufacturing it. We finish it off in the computer. But it has to function as if it were real."
	"The animation on the character of Alita is made up of a couple of things," Baneharn elaborates. "Broadly, we capture the body of Rosa for Alita. We do a [motion] capture process where the skeletal data in real-world space is extracted to digital [using performance capture suits and reference dots]. At that point, we start to make decisions about what it is that's specific about the performance that we want to hang onto, and what specifically we would augment or enhance. On the facial side of things, we do a facial capture, but then we animate all of the layers on top of it. That gives us probably seventy percent of the performance in the basic capture. But the devil is in the details. The last ten or fifteen percent of the credibility of' the performance is the most important. So that's the stuff you're striving for — how to make it come to life, how to make it feel real, and make it feel as if Alita has a soul."
	To capture Rosa Salazar's facial performance, the actress wears a "head-cam" helmet with a rig that has a camera mounted on it, pointed at her face, where makeup has placed a grid of dots on Salazar's skin. "The camera captures what happens at the skin-surface level, that Matrix of movement," Baneham explains. "Then we willfully re-engineer the actor's facial movements [in the computer] to create a nuanced performance from Alita that feels real to the viewer."
	There is a continual back and forth, dialogue between the filmmakers, the VFX artists and Weta. "The process of going from concept art to finished visual effects is continually evolving," Jon Landau notes. "We don't turn [the art] over and say to Weta, 'This is exactly what we want.' They have to now build it. When we see Alita's face come to life, there are little differences in a static pose where you're not seeing her talk, you're not seeing the full range of expressions," Landau continues. "Those moments are when we really look at the detail of, 'What is the mouth doing?' Richie Baneham is helpful in all of this. We continually evolve it, not just as a design, but also sometimes on a scene by scene basis, so that we can get to that range of emotion in Alita that Rosa gave us in her performance. We're continually asking. 'What modifications do we need to make to get her performance coming across?'"
	For non-human objects, such as the obviously mechanical cyborg body parts, "From a design standpoint, there is a utilitarian nature to the mechanics that need to be understood, and we worked with [visual consultants] Ben Procter and Dylan Cole on those to better understand what the intended look is," Baneham adds. "From there, we extrapolate what we think we can achieve based on the plates. We mark it in with Mike Cousins, the head of animation at Weta, and then we put it up for review and try to get a consensus from all of the key artistic players as to whether or not it achieves what they want."
	Procter and Cole "have been heavily involved in the VFX reviews, looking at the assets. In my case, the assets of all the cyborgs as they've been developed, both as shapes and designs," Procter continues. "Not the animation — Richie Baneham oversees the performance animation. But as far as the mechanical aesthetics of how something moves, we all critique it together with Robert in his reviews. For example, the animators' may come to us and say, We need [this cyborg character's] arm to extend really way up high, but look what happens to the piston in the scapular area when the character performs that action: it loses connection. What are we going to do?' Then we agree, 'We're going to add a pivot here, we're going to put a triple telescope on that piston instead of a double, based on the function they need.' Overall, the aesthetic is of trying to solve things.
	"The way that the characters arc ribbed reflects that, with pistons and motor housings," Procter concludes. "They tend to behave within the laws of how they should operate. If it looks real, and it feels real, we've given the right aesthetic intention to the audience."
	The process on Alita involved an uncommonly high level of communication between departments, according to Procter, with the design artists, 3D artists, animators and filmmakers all conferring with one another. "Typical movie, you supply artwork — 'This is the intention'—and then people on the other end, usually without any communication with you, have to figure out how to make it work," he says. "In my experience, its a little unusual that the Alita visual consultants are involved in the way that we are, looking at dailies."