Brian Rausch, manager, motion capture and cinematics group and Aaron McFarland, editor/compositor in the cinematic solutions group, work hand-in-hand to provide the high quality visuals that accompany many of SCEA's games. With a huge motion capture studio, these guys are responsible for putting the players into SCEA's sports lineup, the dragon riders into Lair, and the soldiers into the upcoming PS3 sequel to Killzone.
What tools do you use [for motion capture and animation post-processing]?
Brian Rausch: To acquire on-stage, we use Vicon IQ, Vicon Diva, Motion Builder, and our end product 90% of the time is Maya.
Aaron McFarland: [regarding cinematics and video services] We're a Mac editing house. We are almost entirely on Final Cut Pro, and then Shake for our compositing. On the back wall we have high-end editing suites and compositing suites in each bay. It's kind of a separate little Macintosh pod in an otherwise PC world.
BR: This is our lighting and rendering area. This room's pretty malleable. When we staff up for production, we'll bloat that area. We're right at the tail end of Lair production, so our lighting team is the largest team right now, and they're in this room. When we start our next production, we'll move the animators back in.
BR: This is our online room.
AM: We do an offline edit, and then at the very tail end as we go up to HD, we have a PlayStation 3 development kit all wired through to every possible file format that a marketing guy, PR guy, or anybody could request. Any team within Sony who wants anything in any format can come to us and we can either capture it ourselves or take what they can give us. We cut a lot of the trailers you'll see.
BR: This is the control room for the machine room. We bring in objects and use these surface markers to realign textures and realign the model once we've acquired it with the rig. Our target texture size is always about 4k. Ultimately, this room was specifically designed so that we could bounce light off the walls, and make everything incredibly flat. If I were to bring on the other lights, what will happen is that there will be no shadowing on an object, because all the lights bounce back from all angles. Then we're able to control the lights and the darks inside of the game engine, versus having any prebaked shadowing, or worse, prebaked shadowing where we have to have an artist go in and pull that shadow out of the texture. We built this room to specifically avoid issues like that.
BR: This is our scanning room here.
What are the dimensions of this place?
BR: 120 feet by 120 feet by 35 feet high. The trusses in the ceiling are capable of holding over 5,000 pounds. We can connect them together if we need to go above that, so we could pretty much shock-lift an elephant off the ground if we ever need to. The floor was ripped out and laser-leveled so that it was about as flat as a human can make it. The point of that is to keep the data drifts out.
Even in our smaller space, we didn't do anything with the concrete, and you'd have waves in the concrete and have an angle. We don't have that issue here. We're using Vicon MX-40s. We have 88 of them in the room. We don't use them all concurrently in the same zone normally. We can connect them together, but that's just a tremendous amount of overkill.
BR: We can talk about the difference between the two volumes. With the larger volume over here, we capture full body motion. It's gross hand motion but no finger motion, and gross head motion but no face motion. Over on this side, this is our integrated volume. We'll capture full body, face, and fingers simultaneously. We have sound-controlled the environment, and all of these over here will surround certain areas as well to control the audio even more.
BR: For our NBA game, we acquire audio as we're capturing, so we capture the entire production all at once. We don't do separate audio tracks or separate VO. It's all acquired simultaneously. There's something about when an actor acts, versus when he stands inside a control room and tries to deliver lines. There's something so much more natural and flowing when an actor's acting and we capture his voice at the same time.
Do you have to do a lot more takes that way, though?
BR: Yes, and the only reason we need to do more takes is because we live in San Diego and there's an Air Force base close by. This room is not soundproof; it's sound-controlled. So when we get jet fighters going over, it causes a little noise in here.
AM: Actually, it's the Marines. It's helicopters and F-18s.
BR: I'll take you to the control station here. This is the Sony Anycast system. Essentially, this is a real-time view of all the marker data and connections. We've got multiple reference video. All the cameras are controllable and zoomable from here. One of the interesting things about this is that I was over in the UK last week, and they were actually able to sit at their desk and direct a capture from the UK in real-time to our stage.
AM: We can direct like a live television show. We can go from up to six cameras, remote controlled on the camera to zoom in on the actual action for an individual shot. We can broadcast that over Macintosh's iChat. Everybody's on headsets, so they can talk to a director out in the space who can come back over and have a face-to-face teleconference in the middle of the mo-cap session for them to put their two cents in. In the meantime, other folks on the team can sit back and watch a "making of" for the motion capture for their game, like a television show, live as it happens.
BR: By actually being able to control the camera views, people are able to stay focused quite a bit longer. It's something we're capable of doing for up to eight hours a day.
AM: All the video of all the cameras can be captured so that when we clean up data or whatever later, we can go back to the reference video and see if there was some noise that got in there. We can see what really happened, and we can do motion editing based on this reference video.