Posted by Craig Barr, 15 October 2012 7:34 pm
Steve Williams is presenting a very interesting talk at AU 2012, entitled “Staring into the Abyss: The Evolution of Computer Graphics in Entertainment”. If you are attending, be sure to register for his class by clicking on the link above. More information around Autodesk University can be found here.
Simply put, Steve “Spaz” Williams is a legendary pioneer in the world of computer graphics in feature film production. Steve’s career spans several landmark CG moments in film such as: the pseudopod in The Abyss (the first “soft surface” CG character), the liquid-metal T1000 in Terminator 2 (the first CG main character), the infamous dinosaurs of Jurassic Park (first digital animals), and, the contortion of Jim Carrey’s face in The Mask, amongst others. The Abyss, Terminator 2 and Jurassic Park each took home the Oscar for Best Visual Effects in their respective years. Steve himself was nominated for an Oscar for The Mask. Furthermore, all are projects that represent important leaps in CG, in terms of developing the techniques and pushing the boundaries of what was thought possible by the all mighty pixel. Aside from Steve’s work on these milestone movies, this is THE man that, on his own time, created an animation test that convinced Steven Spielberg and George Lucas that CG was the way to go for Jurassic Park.
With his extensive feature film experience, Steve has since focused most of his attention on commercial direction. Prestigious honours include: being selected for the Saatchi & Saatchi New Directors Showcase at Cannes, receiving multiple Clio Awards for Electronic Arts, as well as for the Blockbuster “Carl and Ray” campaign.
During his studies in Animation at Sheridan College, Steve spent his summers working at the University of Toronto’s OISE Computer Lab. Upon graduation from the Classical Animation program at Sheridan (the days before there was a Computer Animation track), Steve went to work at Alias Research (yes, some Autodesk history there) in Toronto. It was here that Steve became proficient in Alias, a software package that would completely alter the landscape for design, graphics, animation and visual effects in feature film. In 1989, Steve joined the Computer Graphics Department at ILM.
It was movies like The Abyss, T2 and, certainly, Jurassic Park that got me into this field to begin with and it was a great pleasure to have a chat with him about how much, and how little, things have changed since those early days.
For the last few years I have been directing commercials all over the world. I have directed well over 300 commercials. I also teach a class on Directing for Film and Animation at the Academy of Art in San Francisco. I also teach a lot of science in that class too. Yeah, I know, not as “glorious” as the old days! I spend a lot of my free-time building in wood and metal. (you can check out some of Steve's work here and on Vimeo here).
I was at Alias Research back in 1987 doing demos and I became a bit of a hot-shot demo guy showing-off Alias. When I had started at Alias they had just introduced version 2. I remember I had to demo at NAB in Austin and the damn machine kept on crashing. We were on an SGI 3130 and the configuration was buggy and it was embarrassing as hell. You’d be in the middle of a demo and the f!*#@*g machine would hang up!
Yep, that’s too funny. I know the pain.
I was the first Alias person at ILM in 1988. I remember very clearly that I was using Alias v2.1. We had a staff of about 7 in the department. I was the first animator and the first Canadian in the Computer Graphics Department. I was brought in to do the first demo of the pseudopod for The Abyss for James Cameron. Then they said, “Oh, we’re going to keep this guy”. Mark Dippé and myself built the pseudopod. That was 18 scenes…92 seconds… and that kinda really fell on the radar for CG. It also brought Cameron back to us for Terminator 2. I was Animation Supervisor on T2. Then we heard about this Jurassic Park idea. I became Chief Animator on JP and we took all the data and ported across to Softimage because of its Inverse Kinematics system. Those 3 films kinda laid down the tracks for the entire industry, AND…at the basic level, not much has changed. I was running Alias on an SGI GT60 and GT80…these were the first machines to run Alias at ILM. I can’t remember what the cost of those machines were, but I know it was insane.
You had two options for animating in the early versions of Alias. You could animate an object with Translational, Rotational, Scalar (TRS) data, or you could animate control vertices. Alias was very good for animating fluid motion via B-Splines. You could easily visualize the behaviour in the viewport. The key was to be able to animate the spine through the Pseudopod. This spine that ran along the pseudopod, carried the basic animation, but then we had a series of control vertices, like slices, that we would further refine as we went. I would take a spline and animate the CVs along a path and make sure to maintain the exact distance between the CVs. Then I would animate separate slices or rings of CVs along the spline. These rings would be orthogonal, or flat. Then I could animate what they were doing, like they were undulating for example. Then we had a program and it would take each ring and lay them down along the corresponding CV at 90 degree normal, then high-res them, and then run them through SOCK (more on that later) and it would create one complete object. We had some control over the resolution of the surface and we would lay down individual sine waves distribution points along the entire thing that would create the rippling effect. We had a one-bounce ray trace in Renderman, which would create the optical disturbance in the background.
A documentary special highlighting Steve in 1992
There were about 5 of us in the department in 1988. Mark Dippé did the code for the pseudopod and I animated most of the shots.
Absolutely, back then we didn’t even have all the pieces or the parts to do what we wanted. We essentially had to manufacture all the parts.
Cameron came to us with an idea that there was no solution for: a realistic, alien, pseudopod of water. Those solutions had to be built with early, off the shelf software. We had to build a lot of custom tools to help us accomplish what we needed. During Terminator 2 we had another software problem that was solved by another guy from Alias, an Engineer named Angus Poon, who came down to ILM and wrote a program that we called “SOCK” that was responsible for maintaining C2 continuity across these 4-sided, B-Spline patches. My partner in crime then, Mark Dippé, knew the higher level of math that was required and he wrote a lot of good tools.
The face data for the placement on the end of the pseudopod was a serious accomplishment. It’s taken for granted today, but using facial scan data and stitching it on to the end of this living tube of water, with b-spline patches, was a real challenge.
I’d have to say our ChanMath program. When you were animating multiple channels of control vertices, you ended up having multiple pivot points in space. This inhibited control when choosing to "group" vertices with existing animation data into a new hierarchy. ChanMath would collapse all the control vertices to 0,0,0. I personally would use this all the time when animating the "death sequence" (the end lava scene) in T2.
For the T-Rex muscle animation we used animated, low-res bladders in the legs and keyed the Y-axis bouncing up and down. Then the CVs on the high-res skin were instructed to listen to that bouncing bladder on a weighted scale from zero to 100%.
The dinos were all built with b-splines. They were put through our program SOCK to align and close the surfaces, and then we had these guys that were the “Envelopers”. It was very, very tedious and these guys would constantly type and run tests to check their enveloping on the skins. These were the guys that would blow out their hands and arms with Carpal Tunnel Syndrome!
I built the first bones for the Rex and, later on, the entire creature. Yet another Canadian, Eric Armstrong, built the Raptor.
A still from Steve's T-Rex skeleton run-cycle that changed CG & film forever.
Figuring out the joint placement with Softimage units (scan courtesy of S. Williams)
I was Chief Model Builder and Animator on Abyss, T2, and JP (T-Rex). I built, rigged and animated all of those things. During T2, we came up with a way of building teams to create shots that used a lot of the same techniques and data.
Same platforms as the projects before hand (The Abyss, T2, Jurassic Park, Jumanji, Eraser, ..) – Alias on the front-end for modeling and Softimage for animation. The IK system in Softimage was superior. There was a little bit of blend-shaping going on for The Mask. For the most part we used a library of shapes that we would look through. Cary Phillips wrote a system called “Caricature” (“Cari” for short; won a Technical Achievement Award in 1998). This library allowed us to access all of the exaggerated shapes and extreme expressions for Jim Carrey's face. We also used a bit of spline-animation mixed with blend-shapes that followed those splines when the Mask would wrap around his head. One of the biggest challenges in Spawn was his cape. We had chains of bones driving the basic, overall structure and a fluid dynamics algorithm running over the fabric.
We were using a very early version of what became Photoshop as early as The Abyss with no ability to use anything like UV coordinates. For T2 we were using a version of UV mapping. We used coded and texture-based shaders for Renderman. Everything had to be scripted out into RIB files to go off to Renderman.
There was no rigging or anything like that. You had hierarchies of B-Spline patches on a rudimentary forward kinematic system. It was very crude but it did the job and it allowed me to figure out a way to build the T1000. We didn’t have the ability to project a curve onto a surface, we were stuck in B-Splines. B-Splines allowed for an easier way to shape surface patches. We also had a version of Steven Coon’s patching system. This was way before NURBS and Subdivision Surfaces wasn’t even on the radar. We didn’t have access to things like Mudbox or any of these tools to sculpt and bring up detail. We did have source code to Renderman because we had a good relationship with Pixar. But, you know what? I had animation under my belt, I was a good animator and I thought I’m going to make this f@@@er stand on end. And I did.
I was not invited to that meeting, unfortunately. I heard that, apparently, after my T-Rex bones test (I set up low-res bones in Softimage and animated and rendered out a test) that Lucas was quite emotional. He was so blown away but what he was seeing.
No I don’t think so at all. If anything I feel for the artists today. The problem is that Hollywood puts more pressure on the artists because the tools are always becoming better. Clients expect a Cadillac for the price of a Chevette.
"Carl and Ray" - Blockbuster
That transition was very fun. Directors usually go the opposite direction - namely, from commercials to movies. I love shooting live action comedy, though the work I get is as a reflection of the work I've done while at ILM, really.
Capital One "Mailbox"
Great in many ways, horrible in others. I am not cut out to deal with certain types of individuals that run large Hollywood studios. I have checkered teeth, I smell like a Slim Jim and enjoy crushing nerdy executive hands. Not a good combination for getting your foot in the door in "Hollywood". Look what Hollywood has done to computer graphics .... turned her into a slutty sister!
Lucas wanted extra people and activity going on in the background and asked me to take care of it. I actually got to Direct myself. I just set up a camera to shoot 50% live action and 50% background, with a green screen. The picture below shows myself and Paul Hunt walking through Mos Eisley.
Politics and internal debates aside, Jurassic was great. The Abyss was so under the radar, no one had a clue what we were up too and we were left alone. It wasn't until after the success of The Abyss, that this CG thing took off and all hell broke lose for Terminator 2. Everything suddenly became much bigger, even though it was still Mark Dippé and I who set up all the teams and tech.
The resurrection of deceased actors, whether people like it or not. Here’s a link to a paper I helped out with concerning the subject: www.law.berkeley.edu/journals/btlj/articles/vol8/Beard.pdf
Please only report comments that are spam or abusive.