Ed Ulbrich is the President of the Commercial Division of esteemed visual effects house Digital Domain, and a highly accomplished visual effects supervisor and innovator. Having worked previously with David Fincher on Fight Club and Zodiac, he once again collaborated with the director to bring to the screen an Oscar-winning milestone in visual effects – the first completely convincing, ‘stare-able’ digital humans, in Fincher’s The Curious Case Of Benjamin Button.
If you’re not entirely sure which parts of the film were CGI, check out Ed’s jaw-dropping 18-minute presentation at TED, in order to appreciate just how much ground was broken to portray the story of a man who ages backwards, and to allow principal Brad Pitt to truly act out all the stages of his character’s extraordinary life…
What can you tell us about the attempt with Ron Howard in the early nineties to see if Benjamin Button‘s effects could be achieved with the technology of the day?
Well, we didn’t try anything, to be honest – it was more like theory. It was the philosophical approach of ‘How do you tackle this problem?’, and you break it down into smaller pieces and say ‘I think we can do this, and that and that…’. We managed to identify the ‘unknowns’, anyway.
At that point it was the early days of digital stunts, which we used to call ‘face-paste’ and head-replacements. With the Cyberscan technology that was coming out, you could take a scan of a 3D object like the human head. These were the early days of gluing digital heads onto stuntmen that looked like the actor.
We were noodling round ideas, like with James Cameron’s Titanic and another film he was doing at the time, where we were thinking about – ‘Wow!’ – doing complete digital stunts. [laughs] Again that would only hold up in the very wide shots.
So you were thinking of a multi-stage approach – partial CGI and prosthetics, and so on?
Yes, but it wasn’t just sticking a static head and a frozen expression on an action shot in the context of a giant stunt, where the camera’s moving and there’s explosions and you just quick-cut to it. This was like…the guy’s got to hold up in close-up and talk and carry out a performance. And that was way way out of the limits of what had ever been done before.
Marker-based motion capture was just coming on the scene, but we realised that so much technology would need to be developed from scratch. Could we have done it? Sure, but…
For two billion dollars.
Exactly. It tips the scales of feasibility. It wasn’t a reasonable proposition. When we did it just now, it was still very difficult, very expensive and it took a long time, and the more you do it, the faster, better and cheaper it becomes. Back then [laughs], it was the early dawn of time, as it relates to this technology; crawling out of the primordial ooze. It was too early, and would have cost a shocking fortune and taken many, many years, maybe.Those of us who dabble casually in CGI are curious as to how an ‘analogue’ phosphorous layer of make-up can generate consistent polygons, in the process whereby Brad Pitt does his motion capture without those stick-on balls we’re all so familiar with from DVD extras…
With marker-based capture, it’s not actually capturing polygons. It’s just capturing key positions on the face that correspond to places that give you the largest muscle-motion. If you do the body, you don’t capture between the points, because it doesn’t give you any useful information, really. With the human face you have the cheeks, the eyebrows, all the areas that you can animate that are so significant. But that’s not where the geometry comes from, where the polygons come from. The polygon comes from either a cyberscan of the person, where you put the person in what looks like a large phone booth and they get their head and their body scanned, or you do a life cast of them in plaster and you scan that. And from that scan, you then get the polygons.So the captured performance is actually distorting this rigid digital maquette?
Correct. When you put marker-based motion onto those polygons, it’s effectively like stretching a rubber mask. The motion of the skin over muscle and bone is not what you’re getting. You’re getting points which you target onto the geometry of your scan of the performer, and it’s pretty freaky-looking. So we found out that we really needed to know what was going on between the markers. We needed the wrinkles and the creases, and to understand the behaviour of the skin over the muscle and the bone – we would never get that with the marker-based stuff. When we first got involved with this company called Mova up in the San Francisco Bay area, it was the early days of this technology and it was clear that they were on to something pretty substantial.
What you were able to get was the idea of creating sequential geometry, so instead of creating data of points in space, we could, in fact, create a unique 3D model for each frame. So the idea is that you can perform it and it makes a 3D CG model of you, frame-by-frame.
Instead of having hundreds of markers on the face, we would stipple on this paint…it’s actually not paint, it’s make-up – you naturalise it and it looks invisible. But when you hit it with UV light, it glows like a phosphorescent green. When you stipple it on with a sponge, what you end up with is a random pattern of probably millions of little markers or dots.
So then you step in front of all these HD cameras, and they’re all seeing you from a different point of view, and all the fields of view are overlapping and pointing to the same subject. Then the 3D software behind it can use these points of view to stitch together a 3D database.Can you tell us something of the challenge of making a human eye look right in Benjamin Button? Was it a question first off of observation and trying to understand even what it was that you were trying to understand about the eye…?
First of all, you go back to where others have gone before, and whether there are things of value that we want to repurpose or use. You have to look at different techniques and technologies and really free your mind, which is hard, because you can fall into conventions very easily. You have to try and be clinical and objective, and ask ‘Can we do better than the very best that’s ever been done with this particular technique or methodology?’.
And that’s where we ruled out marker-based motion capture, because we’ll never make it look any better than the best person who’s ever done it with the most money and the most time – that’s the limit of that. That ain’t gonna work.
With the eyes, we looked at what was missing, what didn’t work in the attempts with the other characters [from previous films]. The thing that we kept coming back to, looking at random shots from various movies that we were studying, was that there’s no ‘life’ in those eyes, telegraphing the actor that’s behind them, like that old adage about eyes being the window to the soul…
Flip over to us really doing in-depth study of the Ekman research from the early 1970s, the Facial Action Coding System…there’s a lot of information there about communication, about muscles and about the human face and the human response to emotional stimuli…all those things which affects the eyes and the position of the eyes and the muscles around the eyes. You just had to really look at the eyes both anatomically and regarding how they respond to emotional stimuli.And that was what was missing. The eyes are very complex systems, how they wince and crease and where they look; if you’re pondering something, your eyes will look up to the left, but if you’re bashful or you have a moment of embarrassment, your eyes will look down to the right. All those different kinds of things. We were studying that behavioural thing that will happen, that involuntary response of the brain telling the muscles to move and then fire. Through the research, you realise that these are the things that need to happen to bring life to it.
So we’d study what Brad would do in order to build a muscle-system that could control the eyes and match what he’s doing. And that was kind of mission-critical. It was also clear from many of the films we studied, perhaps because they’re so difficult to do, that there was an attempt to not feature the eyes. So the idea from the very beginning was that if we could keep people looking in Brad’s eyes and show him behind there, then we win, because we actually do look in people’s eyes when they talk. And our eyes dart from eye to eye as we speak.
So to get an authentic and genuine and sincere and earnest performance through all this technology, it had to telegraph through the eyes. So that part was so tedious…
But Brad couldn’t put make-up on his eyes…
Right, so we had to create an entire other system. The thing that we did get from doing the FACS approach, where we were scanning Brad doing these different poses, is that we did get a lot of what happens around the eyes. We didn’t get the actual eyeballs themselves, though. But we took that as the basis and went into it and added more controls and more detail.
It’s a landmark achievement as far as I’m concerned. What’s the next big landmark for you in visual effects – the stuff that you just can’t do now?
It’s interesting, because we’ve always had one of these things identified as the ‘holy grail’. We still have a long way to go. We’re doing our next project now, another major film that’s taking the Benjamin Button technology and doing version 2.0. We’re very proud of the work on Benjamin Button, but I can look back on it now and go ‘oof’, regarding the shortcomings and some of the issues. So we’re now meaningfully improving on that as a basis and taking that forward.
I think we’re in the early days of human animation, completely and utterly believable human characters. I think there are enormous applications for that. When it starts to become more approachable and attemptable to use it, you’re going to see it start popping up in advertising and videogames. That has huge implications, I think, in other media. Think about celebrity endorsements for brands, your mega sports-star or your movie star. They’re always available.
What about the ethical issue with the possibility of reviving dead movie stars digitally?
Well, there are certain ethical issues, but approaching the idea of doing someone who is deceased is wildly more complex. I get asked a lot, like ‘Now you can bring back John Wayne or Marilyn Monroe…?’. No, actually.
Because they’re not here to give you the information you need, in terms of capture.Exactly. They’re not there to actually give us the performance. The whole idea of FACS, this research that Dr. Ekman did, is remarkable; each of us have common anatomical attributes, and we have common similar responses, but they’re all somewhat different. [Ekman] has mapped out and catalogued the human face, but the size and shape of our muscles and how far they can constrict and expand…it’s unique to each of us. The groupings and firing of the muscles are similar, but each one of us has our own little signature and tics and idiosyncrasies. Each person moves through these FACS poses in a similar but different way.
Without the benefit of the real person, their real emotional response and the way their muscles really work don’t come through, and so you move back into the realm of interpretation. And once you go into interpretation, it’s a futile proposition.
Did Brad Pitt find it hard to enter into the character in what seems a potentially sterile and isolated filming environment?
No, I’d say quite the opposite. At first we were all worried that this was maybe going to be rather daunting, intimidating or even clinical, because he wasn’t going to be there with the rest of the cast…
David Fincher really says it best – Brad took to it like a duck to water. It was really very natural for him. In a lot of ways it was freeing, because he could try things and experiment with things and he didn’t have to worry. It’s not like a lot of visual effects photography, where the actor comes out of blue screen and they have to hit their eyeline and their mark. Brad would just be Benjamin. He’d watch the performance of the rest of the cast and just try things. It goes very quickly, because you’re not worried about blocking the shot and saying ‘action’ or filming anything – it’s just Brad doing his thing.
Nor did he have to worry about make-up or lighting or anything like that -we were just shooting him au naturel. And from that we have to choose takes which we think best match the rest of the cast.
That was the hard part for David, but I think Brad really enjoyed it, because he got to try a lot of different things very quickly. Waaay faster than any of us thought. I don’t know that that would be the case with every actor. Everybody was saying how the technology was complex and cool, but what we were looking for was the ability of an actor to walk on to a stage, not be with the rest of the cast, look at scenes that had been shot six months prior where someone else is playing the body…and now you’re told ‘Okay – you’re now a seven-year-old boy who looks like an eighty-seven-year-old man. You’ve been taken to a faith-healer at a church revival and he lays hands on you and tells you to get up out of your wheelchair and walk for the first time, and…action!’ [laughs]
To become that character in that moment, in that context, is amazing. And then to do it consistently and have that performance change through the days of him giving it, because he gets older and more capable and wiser, and his character evolves…you just go ‘wow’.
I came away from this with an immense appreciation for actors, and particularly Brad’s ability to do this. I think one of the things that’s most noteworthy is that he could perform this way. And then there was David Fincher’s ability to direct all this – it was like directing layers.
Do you have a reliable pipeline for the use of these techniques in your new project, or is it still evolving?
Oh, it will always evolve. I always tell people there’s no destination in what we do in visual effects and animation. In a lot of cases we don’t even call it visual effects anymore – it’s really just digital production.
So it’s not really the destination, it’s the journey. Once you think you’ve got everything nailed and totally figured out, someone else comes along and just blasts past you with something even better and more efficient. It’s all about innovation. There are some pretty brilliant people out there who will build on or be influenced by that which we’re surrounded by. You see what someone else has done, someone else will have a great idea and take it to the next level.
Already we have evolved beyond what we did for Benjamin Button on this next project, and now we’re working in stereo-3D, so you’re dealing with films that are working with the fourth dimension, and that adds enormous complexity.
Are you excited about working in 3D?
You’re asking me that at a bad time [laughs]. Right now we’re in the middle of it and it’s really hard. So I would say that excitement’s not what I’m feeling right now – it’s pain!
Can I ask what the project is?
So I can ask all I like!
I can’t tell you.
A science-fiction project.
Really can’t….it’s a project we’re really excited about…
Something we’ll be seeing in a year…two years…?
You’ll be seeing it sometime, err…in the future!
Fair enough. Could I ask you what the bottlenecks are now to keeping visual effects evolving? Is it time, research, hardware break-throughs…?Time, money and faith. The enemies of technology in the service of creativity are tradition, complacency and fear. People who fall into that tradition become complacent in the tools that they have, or afraid to pioneer and go beyond. And like that, you never will. It takes partnering with brave film-makers. I would say that David Fincher is probably the most bold, innovative and fearless director that I have ever met, let alone worked with. David is one of the smartest people I’ve ever met. He can go toe-to-toe with some of the most advanced programmers and animators, compositors, cinematographers and technicians, software developers…he holds his own. He makes it his job to understand what everybody else does. He can be fluent in it and exploit it and really use it.
He pushes us. It is amazing how he comes up with these ideas, because he is such a…geek [laughs]! He eats this stuff up. He’ll just go and soak up this knowledge, come back and say to us ‘What if you tried it this way..?’. And you go ‘Wow, actually that’s brilliant’.
So it’s safe to say that David Fincher has kept himself completely up to date since his days at ILM?
Up to date? There are few people who would have attempted this, on Benjamin Button. You’re talking about carrying a major motion picture with the performance of a digital human being. And it’s not an effects movie. There are no aliens and spaceships and explosions. This is a period piece, a story about a man’s life, and that performance is mission-critical. If it doesn’t work, the movie fails.
I didn’t realise that the character used an entirely digital head until I saw the TED presentation a couple of hours ago. I didn’t know how it was done at the time that I saw the film – I figured a mix of prosthetics, alternative actors and CGI…so you got me! You must be very proud both of the work and the Oscar.
I am, but really appreciate you saying that. It’s weird, in our world, that the highest praise is when people really have no idea what we do…