Filmed in Sydney and released 20 years ago this week in 1998, Alex Proyas’ Dark City surprised many by instantly becoming a neo-noir sci-fi classic. Fans of Proyas’ earlier works, including The Crow, were perhaps not surprised at the director’s adept skills in crafting a film around a dark sun-less world run by a mysterious group known as the ‘Strangers’.
Helping Proyas to achieve a distinctive look and feel to Dark City was Peter Doyle. Doyle is best known as championing the digital colour grading on Peter Jackson’s The Lord of the Rings trilogy, as well as lending his colorist skills on the Harry Potter films. Now he’s a supervising visual colorist at Technicolor, with recent credits on films such as Fantastic Beasts and Where to Find Them, Darkest Hour and Paddington 2.
But before Lord of the Rings and all those other blockbuster films, Doyle made his mark at Dfilm in Sydney, a film production services lab formed in 1995 that Doyle joined to help establish a wider digital post-production ambit. One of Dfilm’s first prominent gigs in that regard would be Alex Proyas’ Dark City (the company would later go on to work on The Matrix, also filmed in Sydney).
On Dark City’s 20th anniversary, vfxblog asked Doyle, who is credited as visual effects creative director on the film, about his time on the project, which came right as traditional film workflows were truly transitioning to digital, and as Australia’s own visual effects industry was just ramping up.
vfxblog: What were you working on right before Dark City?
Peter Doyle: I was actually working over in Massachusetts with Joel Hynek and John Gaeta on a couple of U.S films. They were Eraser, the Arnold Schwarzenegger film, and we were doing some of the initial development work on What Dreams May Come with director Vincent Ward. So I was working over there and I had a call from Atlab in Australia which owned Ddilm, and they wanted to expand into visual effects. I came back to Sydney and then we met up with Andrew Mason, the producer of Dark City. We put together a proposal where we could service the film, basically.
vfxblog: What were your first memories of seeing the script, or any kind of footage, or talking about the style of Dark City?
Peter Doyle: Well, it was very exciting. Once we read the script, it became clear it had a very strong image. I know Alex had put together, with David Goyer, a good story, and the concept was very strong. As an overall film, it was a solid world that one could really get involved with. The whole idea that the entire film is set at night, except for the last shot. And then with Dariusz Wolski, the DP, his photography was looking really good, and the production design was very strong. It was quite exciting. Certainly fresh, in one way. It was an interesting combination of film noir and German expressionism and a lot of influences that, to Alex’s credit, pulled together into quite a unified world that, I think, held up.
vfxblog: What became your role in relation to this film? Was it pulling a team together to do the effects or were you doing other things at Dfilm?
Peter Doyle: I guess it was like a hybrid visual effects producer/supervisor kind of role. It was only because of the experience I had working in U.S. facilities. I was still working on a consulting basis with Kodak, at the time, on their Cineon digital intermediate system, which was being developed in Victoria, Australia and in Rochester in the States.
My role was to basically put the team together and work out the pipelines and become more of a creative director rather than a visual effects supervisor. Mara Bryan was the visual effects supervisor. It was an interesting film in that it was a hybrid of model and miniatures work and practical effects and digital. My role was really to put the team together and work out, with Mara, the best breakdown – ‘Okay. We could do that with CG,’ or, ‘That’s going to work better with models.’
vfxblog: Back then, what were some of the challenges of shooting on film, getting it digitised and working on it? Can you talk about that, especially in relation to Cineon and anything else that Dfilm was doing?
Peter Doyle: Certainly, at the time, it was reasonably new technology. Dfilm, being part of Atlab, had quite the store of intellectual property, in terms of colour management and so forth, so we worked extensively with Kodak. We put on site a Kodak scanner, which was actually the first in the southern hemisphere, I seem to remember, and we also put in a laser recorder, which was one of seven that had ever been built. It was really to get a team together and put together a classic scanning and recording pipeline, and then working with the lab and getting the lab used to the idea of printing to and developing to much greater tolerances than had been used historically.
The other thing that’s not particularly – I guess it’s not really considered very sexy technology – but really, it’s got to be remembered that this was at a time when you shoot on film, you have lab rolls, they are cut physically and spliced together, and the visual effects are spliced into this cut as well. In some ways, you don’t get a second chance, because you’re dealing with the cut neg, so it was also putting together a pipeline that actually allowed for that; working with the neg matches at the time, and putting in key code systems and tracking, and that whole side of things that, sadly, now is a completely lost skillset. I don’t think it’s even possible to put that together these days.
That was, I wouldn’t say a challenge. Certainly, in Sydney, there were some great neg cutters and the lab was certainly up to spec, so it was really just a question of just putting in that pipeline and just getting everyone used to that way of working.
vfxblog: What about when it came to colour grading? What were the tools you were using back then? I remember one particular scene which was that sunrise – how were you grading that, for instance?
Peter Doyle: That film was very classic, and Dariusz shot neg and we worked with, at the time, Arthur Cambridge at Atlab in Sydney, who was Australia’s leading colour timer. And that was classic film grading, so it really is printer points, moving dichroics around, and really just RGB. Just all the discipline that goes with that.
Then that sunrise, that final sequence, you could say is probably one of the first DIs ever done, as such. That was very much a hybrid approach, in that, at the time, we were using what we call a CRT film recorder. That was very much a hybrid approach, though. We scanned the film negative in and got that into digital, which was a classic Cineon file format. Then we used, at the time, the Cineon compositing package as the base engine for rendering and colour management. Then within that, I had just some custom code written to allow for grading, as such, but it was approached much more as a series of visual effects, rather than a grade, because we were stripping out skies and putting in full sky replacements to give that hyper-real effect.
To get that intensity on an actual film print, I approached the recorder process using a trick that’s sometimes used in animation at the time, which was to double expose the actual digital recording negative, so that we were chemically able to build up this intensity, so that then it held up against the original photography.
I seem to remember that actually worked quite well. It just took a little bit of time to get everything calibrated. Certainly, it’s a very different approach to how you would go about grading a film these days, but that was the basic concept at the time; was to treat each shot as a visual effect. But one of the advantages of that particular software, at the time, was that you were able to carry multiple shots at the time.
We had what, I guess you would nowadays call, a timeline, as such, but it certainly was not fast. I do seem to remember that we would really have to schedule the rendering and the RAM usage, because this stuff at the time was just so intense, in terms of computation times, that I do seem to remember we would actually physically move RAM from our different computers, which was actually SGI at the time, to actually be able to make these things work. I seem to remember we were up to 60 and 70 hours of rendering, just for a colour grade.
vfxblog: What do you remember about some of the individual visual effects shots and how they were achieved?
Peter Doyle: Well, politically, there’s one philosophy that might be interesting to note, in that, when I put the team together, we had a mutual agreement to try and experiment. You could say it was almost naïve in some ways, or optimistic. The group of us was really small. I think, maximum, it was about 20 people. But the core group, we all agreed that we would try and keep the hierarchy flat, rather than a supervisor and then a lead and then a junior. Rather than go through that concept, we all said it might be really interesting to, just as a team, work as a flat level. It was much more of a socialist approach to things, in terms of shared responsibilities, it just meant that each person could do what they were really strong at, but they were in charge of the complete shot.
For instance, say, the whole water rendering and so forth. We had Dan Kaufman, who we brought out from the States, who was very strong with particle animation. He wrote a water simulation programme using Houdini or Prisms software, at the time, and moulded that in with an actual live action model shoot.
That was one approach, and then the second approach overall was to, wherever possible, shoot using actual miniatures and live action models and to augment that with CG or, in fact, morphing. A lot of the tuning of the buildings were, in fact, miniatures that were, in fact, morphed using a 2 1/2D package. That was really driven by Val Wardlaw, who we brought out from the UK, and she hacked together a 2 1/2D morphing package, again in Houdini or Prisms. That seemed to work quite well.
The tuning zaps and the memory things were interesting, in that we approached that as a hybrid approach and created an A- and B-roll and, in fact, used an optical printer to marry the different layers together, because it just would’ve, I think at the time, blown the company up to try and do that digitally. I think we would’ve shut Sydney down, in terms of trying to render that thing, so we used an optical printer to marry that together. That was interesting because that really gave the whole thing a really filmic and … an aesthetic or an image that I think was appropriate for the film. Faded postcards and that kind of world.
Then Jon Thum had some ideas about the fights where they’re using memories to fight each other. He put together, I think in Matador – an old package that was pretty good to do this kind of displacement mapping. Again, a lot of tricks that now you would probably not attempt to do it as cheap, as such. You would sit down and really build the thing for real in 3D, or whatever, but at the time, it was really trying to augment what was there. It was an aesthetic that was appropriate for the film, I think.
vfxblog: What were the ways that Alex Proyas and other filmmakers would review work that you do? Would you always do film-outs? Would they simply come and watch it on monitors?
Peter Doyle: That was one of the pain points, in that, in terms of the facility that we built, we built a model that allowed for realtime playback on colour-calibrated monitors. First and second passes were done on monitors just to make sure it was all in the ballpark, and then after that, yep, it was old school film-outs, prints, and then reviewing it in a cinema, which requires quite a lot of discipline. It’s quite expensive, it’s quite time-consuming, but at the time, that really was the only way to do it. But I also feel that actually contributes to the process, because people are seeing the effects as they will be on a piece of film.
vfxblog: The film came out in ’98, and then, clearly, The Matrix came along next with an interestingly similar plot, in some ways. You’d already had that relationship with John Gaeta. Is that how Dfilm came onto The Matrix?
Peter Doyle: Yeah, yeah. I’d known John from work back in Massachusetts, and I do remember sitting with Larry and Andy after they had just been shown Dark City. Because when they came through town with Barrie Osborne, the producer, the film hadn’t quite been released yet, so they’d set-up to have a look, and then everyone just siting around laughing, realising that they’re just about to make Dark City again but called The Matrix instead.
Certainly that relationship was there, and some of the crew had also worked with them before. Sally Goldberg, who contributed a lot of animation and CG work, had also been working with me over at Massachusetts with John Gaeta and the team. At the time, it was a very small industry. I think there were 20 Cineon licences in the world.
vfxblog: Looking back, have you had a chance to watch Dark City over the years, and what are your memories about the film and how it holds up in general?
Peter Doyle: Personally, I’m very proud of the work. I actually really like the film. I think, as always, that there was some studio involvement, in terms of cutting, which is kind of inevitable. I personally feel that wasn’t necessarily the best way to go.
In terms of the work and the team we set up, we really had this almost halcyon concept of a group of people that are all equally skilled and working together, rather than what is nowadays pretty much a very industrialised process, which is necessary. I think, if you take on a major film nowadays, with 2500 digital effects shots, it’s not going to work to have a dozen of your buddies and yourself sit there and go, ‘We’ll get through this.’ It just can’t work.
For me, too, Dark City was an interestingly short period in the arc of visual effects production, or at least digital visual effects production, where one could feel that their contribution was effective and great and that one could have a true involvement in the process. The entire team would, in fact, interact with Alex, as such. It wasn’t people presenting the shots to me and then I would filter what is done and then present to Alex; it was much more egalitarian.
Each of the compositors and CG people would present their shot direct, and it was assumed that they would be able to take the notes and work on that. I think everyone that worked on that film in that model holds that memory as that was the ideal way to be able to do a film. But now, it’s quite an industrial process and one’s contribution is quite different and quite filtered.
In terms of the team and just the back-room stuff, none of us had to kill ourselves. We weren’t having to stay there for four days non-stop. All that kind of madness. It wasn’t exploitive. Frankly, the politics of the workplace was positive. I think everyone felt an ability to contribute creatively in a very positive way that it maybe not necessarily the case these days.