“There was ‘stuff’ on that film. A lot of stuff.”
– Chris Woods, visual effects designer and supervisor, Super Mario Bros.
You might think that in terms of visual effects, 1993 would be a year solely remembered for the groundbreaking CGI and other effects work in Jurassic Park. But, as amazing as that work was, another film that year also made waves in digital visual effects: the unassuming Super Mario Bros., directed by Rocky Morton and Annabel Jankel and starring Bob Hoskins (Mario), John Leguizamo (Luigi), Dennis Hopper (King Koopa) and Samantha Mathis (Princess Daisy).
Although Super Mario is often declared a disastrous adaptation of the popular Nintendo game, the film was a game-changer in two major areas; it was the first film that went through the Kodak Cineon film scanner and film recorder just for visual effects, and the first to use Gary Tregaskis’ Flame compositing software, even before it became a Discreet Logic (now Autodesk) product.
To celebrate 25 years since the release of Super Mario, we go retro with visual effects designer and supervisor Chris Woods, digital animator/compositor Sheena Duggal and software developer/animator Kevin Bjorke to discuss the slightly chaotic production and shoot, the advent of Flame, getting their heads around scanning film, and that time they weren’t meant to see some Jurassic Park VFX dailies.
‘This aint no game’
Chris Woods (visual effects designer and supervisor): There was ‘stuff’ on that film. A lot of stuff – right, left, and centre on Mario from beginning to end. I spent two years on that project, which is a long time. And I lived through various regimes. The people that I was hired with – well, the producers, of course, stayed, because they had the money. But everybody else, with one exception, were gone. And they were just about to start shooting when I joined the team. We saw a complete bloodletting and then re-crewing up and then all the issues.
One of the issues was that, although they had their initial budget – I think it was $28 million, which was, back in ’91, that was a pretty good budget, especially for an indie – they were still having a hard time getting the visual effects to fit in their budget, because visual effects in ’91 were very expensive. It was all done analogue. It was all film opticals and all the old technology, models and all the rest. I came on board to help them with that challenge.
Fred Caruso, a producer on the film, knew me as the digital guy from R/Greenberg Associates, where I had been pushing digital. I was putting together the bits and pieces for the film and somebody said, ‘There’s an Australian gentleman that you ought to talk to. He’s got some software you might be really interested in.’ I said, ‘Okay. Great.’ And I met Gary Tregaskis, who was from Melbourne. Gary had been showing his Flame/Inferno system around, which was originally called Flash. Everybody thought it was really cool, but nobody knew quite what to do with it. Of course, the thing is that film scanning hadn’t quite arrived. But I had seen the prototypes, especially from Kodak. So I said, ‘Wow, I’m going to get together with Kodak.’
I made a deal with Gary, who had no customers at that point. It was before he had joined in partnership with Richard Szalwinski at Discreet Logic. We took what was very much alpha code at that point. And I don’t say this critically at all, Gary would be the first one to admit it wasn’t a product yet. There were a lot of bugs and so it wasn’t production ready. And I knew that. So part of the deal that I made with Gary was that he would sign on to actually be in-studio at a brand new studio that I set up just for that movie – to be on-hand to deal with bug issues regarding capabilities and so forth and new requests, whatever it might be.
Crewing up for a digital world
Chris Woods: The next big task was, okay, so who knows how to use this software? And Gary had a friend named Peter Webb in Australia. [Sadly, Peter Webb passed away last year after a long illness]. I got in touch with Peter and arranged to bring him to Los Angeles. I got a couple of other people also, for example, Sheena Duggal came over from London.
Sheena Duggal (digital animator/compositor): I’d been working in London on high-end print jobs. I got this phone call saying, ‘Do you want to come work on this film in L.A. using this software called Flame?’ and I thought, ‘Wow, I’m probably not going to get that offer again.’ But what I did say was, ‘What’s Flame, and can you tell me about it?’ And they basically said, ‘If you want to see it, go to SIGGRAPH.’ They were doing demos of it at SIGGRAPH, in Chicago, in ’92, so I fly to Chicago and Richard Szalwinski and Simon Mowbray were there giving demos.
I take a look at this software, and I go, ‘Bloody hell, that’s amazing!’ What it could do, it was amazing. You know, it had a DVE where you could translate things, and it had one light in that DVE where you could take a 2D image and displace it using another image displacement tool, and then you could light that, and you could do things that I’d not really seen people do before. And one of the things that I’d been doing in the 2D world, is I’d been using Wavefront to create print imagery by getting some kind of 3D object, and then texturing it with another image, and then composing it in a particular way that no-one else was really doing at that time.
Chris Woods: There were others who came and they were a mix of traditional artists like rotoscopers, who’d never worked on computers. I put together a group, it was about 30 people and it was about half and half and then arranged for them to teach each other so that the traditional rotoscope artists would talk to the digital artists about what they knew from the film side versus what the digital people knew with they styluses and their digitising tablets and so forth.
Sheena Duggal: I mean, I got on a plane to come to L.A., and I didn’t know a single person here, not a soul. I got on a plane to come to L.A. to do a job I’d never done before on software I’d never used, and that didn’t occur to me that it was slightly crazy. To me, it was just a big opportunity.
Chris Woods: We also hired another terrifically talented person named Kevin Bjorke. Kevin was the guy who could write code that was creative code as opposed to system code.
Kevin Bjorke (software developer/animator): I found out that Chris needed somebody to come in and spend a couple of days teaching people about how computer graphics work. That ended up lasting 14 months, which was kind of crazy. I found out that they had almost no equipment, and they only had a couple of people who had done mostly video or TV commercial work, but not really large-scale production and big live-action plates, mixing, computer graphics and doing stuff like that.
Sheena Duggal: Kevin was doing these incredible particle animations, that were incredibly sophisticated for the time. I don’t think anyone had ever done anything quite like that. It took many years, as I recall, before particle systems were really used that way in any sort of commercial manner, because it wasn’t until Side Effects had Houdini that it even was something that people could play with, and even then it was complex. Particle systems are notoriously difficult to control, those types of effects animations, because the particles are all at the mercy of all of these different forces that you’re applying to them, and how long they live, and when they burst, and all of this kind of complication that determine how a particle in a particle system behaves.
Kevin Bjorke: The particle system was called Blizzard. One of the things Blizzard had was the ability to – and later systems did have this, but I was really surprised that it didn’t exist in other commercial systems for a long time – it was the ability to use paint strokes to control the image. You could draw paint strokes and I’d written a special paint program that would record the direction and speed of the brush and so then you could say, ‘Well I want this to go here and this to go here and this to go here,’ and so you could give it a lot of human interacting and up the shape and motions of the particle systems. You could even play that back as animation so you could have someone draw something. Say, I want these particles to kind of go over here and then go down in this whole area, you know, whatever. And you could just do that through painting and then the computer would create a new piece of animation that would follow your lead.
Chris Woods: Kevin did that one shot, amongst a lot of other duties, of Dinohattan that starts on the far side of the river. So you see a landscape of Dinohattan in the distance as a wide shot. And then you pull in, ultimately pulling into the window of the tower where Daisy is, which is live action. But that was all 3D CGI, except for the live action in the window. 100% of it, a 60-second long shot. It was really a tour de force piece that Kevin worked on with also some of the matte artists.
Kevin Bjorke: A lot of the software, I just ended up writing from scratch. It didn’t exist, so I would write scripts to do a lot of different compositing work and I wrote the system that was the multi-plane system where I could like split a picture by depth, so I’d hand-paint the depth line and then move the different parts around to create the illusion of perspective, so I could take an individual still photograph and composite moving vehicle shots over it. I guess I was stupid not to patent that…
A strange shoot
Chris Woods: We were shooting in Wilmington, North Carolina, which at that time was called a right-to-work state, which meant you didn’t have to deal with unions. So you were able to save a lot of money. And we had the new directors, which were of course famously, a husband and wife team. And Fred Caruso said at one point, ‘Well, we’ve been shooting for one week. We have a serious issue. We are one week behind.’ And that, of course, was met with disappointment, to say the least. Because to shoot for a week and be a week behind was pretty amazing. But it’s very hard to make a change on directors. So that wasn’t what was really contemplated.
Instead they got rid of the DP, a guy named Peter Levy. The problem wasn’t him at all. And they hired Dean Semler, who had just won an Academy Award for Dances with Wolves. Anyway, I’m there with Dean and he says, ‘Well, what are we going to do with this? It’s a VFX shot.’ And I’m thinking, ‘Well, you tell me what you want to do.’ And he says, ‘Well, mate, I’ve never done visual effects before, so I’m here to learn from you.’
I was flabbergasted, honestly. It was very thrilling, too. But it was like, holy shit – Dean Semler just said he’s learning from me. I’d better make sure I know what I’m doing here. I’d better make sure I’ve got the right answers.
So they get through all the shooting. The money was gone and then some. And not one visual effect had been shot. Not a single one. And they were all boarded out. In those days it was really important to storyboard. I would argue it’s still important, but in those days it was critically important to storyboard your visual effects rather carefully so that it was clear what was wanted. What you were trying to do, so that a team could understand what the object of their efforts was. The result of that was in as quiet a way as possible, Rocky and Annabel were told, ‘Thanks a lot. We’ve got all that footage in the can.’
They were finished with the film until the next step came, which was to get all the visual effects and second unit shots. I was put in the DGA and shot all of the visual effects plates for the movie over the next weeks. We had editor Mark Goldblatt calling out what seemed to be needed to complete the picture. Then I spent six weeks shooting live action plates and came back to LA where I had already set up the digital studio and people had been working away. But there was so much work that needed to establish how we were even going to approach taking on these shots, because as I’ve already mentioned, it was brand new.
My first time at directing was with Dennis Hopper. I’ll never forget that. It was a night shoot. I was there with the producer Roland Joffé. This was one of the scenes where the worlds were merging and Dennis was turning into a different kind of a monster. He was turning into King Koopa. Roland worked with Dennis for about half an hour, with me very respectfully standing to the side – as I should have done, because Roland was a legendary director and the producer of the film. But when it was all done, Roland said to me, ‘I’ve got to be up first thing in the morning. I’m going to leave you to it.’ It was probably 10 o’clock at night by now, and I said to Dennis, ‘I’m really sorry. You’re going to hate me for this, but we have to throw all that away. These visual effects have got to be shot in a specific way or they’re just not going to work.’ And he was great about it. He said, ‘No problem, kid. You tell me what you want and we’ll try and get it.’ We had multiple layers of prosthetics that had to then be put together and look as if it was smoothly occurring. Organically. Roland may well have gotten better performances out of Dennis at some level but it wasn’t going to work for the effects.
Meanwhile, Bob Hoskins was really, really fabulous. He had a broken hand from something or other. There was a shot where he was coming through the portal where he jumps in to follow Luigi. And yet he had a cast on his hand. I was worried. I wasn’t going to suggest something too physical. I said, ‘Well, what do you think? You’re going through a portal. I have no idea what happens when you go through a portal, but it’s really weird, I’m sure of that. You’re going to come out in this other world in a dimension. You’ve got a lot of acting experience. What the hell do you think you would do? And he said, ‘Well, I’m going to do a flip and come through.’ And I said, ‘But you’ve got a cast on your hand.’ And he said, ‘Oh, don’t worry, mate. I was a carny before I got into this. This is nothing.’
When film was film, but still needed to be scanned
Sheena Duggal: Back then, you didn’t have off-the-shelf ways of really compositing things or scanning things. So what was interesting about this situation was it was the first time Kodak had ever scanned for film. Outside of ILM, who had their own film scanner, and maybe a handful of other people at the time, this just wasn’t being done. You couldn’t even attempt to do digital effects, because you couldn’t scan the film.
Chris Woods: Super Mario Bros. was the first film that went through the Kodak Cineon film scanner and film recorder just for visual effects. It was much too expensive in those days to put a whole film through there. The film was all shot in 35mm, so the visual effects plates were all scanned and it was the first feature film through the Kodak Cineon scanners at Cinesite. In fact, they had said they weren’t planning on being open for real business for a while. Being Kodak and being very careful, they wanted to do testing for a number of months further. I said, ‘Guys, I’ve been going down this tech road for over 10 years. I will be gentle with you. But you’ve got to let us put this movie through the system. I know it’s ready. I know it is.’
Sheena Duggal: On the film I got a digital animator credit, and I also got a colorist credit, because the stuff started coming in, and I was looking at these images and I was going, ‘That looks wrong to me. It’s got no high-end and no low-end. It looks like you’re just looking at a mid-tone image.’ I spent a lot of time colour-correcting, or working on colour. It’s a huge part of what you’re doing when you’re working in the high-end print side of things.
Gary Tregaskis was there at the time, because he was writing code to allow us to do things. I think Gary ran a histogram on the image, and everyone agreed that all the information was in the mid-tones. The next thing I know is I’m being dragged off with the supervisor, and maybe Gary, to this meeting with all the colour scientists from Kodak to try and figure this out. It turns out that no-one had taken into account that they were scanning log-based images, and that Flame was an 8-bit linear software package, and so they needed to write lookup tables, LUTs, that would re-map the data to a display so that we could see the full range of image that we were supposed to be working with, because we couldn’t actually view them in log colourspace. Oh, and by the way, this is technology that ILM had spent years figuring out, right? And there we are just rocking up and going, ‘Oh okay, how do we, what? Well, Sheena says it’s wrong…’
Anyway, they write these 2D lookup tables, and we map the information so that we can apply these lookup tables to display right in Flame. Now, of course, all the math is happening on the images in linear space, not log space. Which continued to be the case for years until they started working with linear image format. So that was so bleeding edge, I can’t even begin to tell you.
Jumping into the fire with Flame
Sheena Duggal: Gary Tregaskis was there with us, and so was the late and lovely Peter Webb. He had worked with Gary on the software. There was no manual, there were no instructions, there was nothing like that. There was just Peter. All the other artists were traditional rotoscope artists for the most part, who had never used computers before in their lives. It was a very interesting and unique situation. Peter was the most experienced because, obviously, he already knew the software, and he was doing a lot of what I would call 2-and-a-half-D matte paintings.
I got assigned all of the stuff where they had to jump through this portal, through this wall, so I was literally there designing how to do that. I just came up with this idea, and you’ve got to remember back in those days you couldn’t save the setup for a file, so if I went into the DVE in Flame and I translated this thing here, and I displaced it with this image, and I had a light, and I had this light at 45 degrees to the surface of the displaced image, I couldn’t save that in a setup file. You had to manually write it all down on a piece of paper. Similarly, if I went into the colour correction, I had to write it all down manually on a piece of paper.
What I would do was take one frame, and I would basically do everything to that one frame, until I got an image that looked really good. Then I’d reverse engineer it across the series of images to figure out what I needed to do. And then Kevin Bjorke, who I worked with really closely on the face coming through the rock wall, he was doing the particle system. I would create this displaced image of a face coming through this rock wall surface they had photographed, and then I would give Kevin that element, and he would generate particles off of that. Quite sophisticated stuff, but no-one was doing that in 1992, with particle systems. It was really incredibly groundbreaking. He would pass that back to me, and I would incorporate that into my composited image.
It was like, ‘Okay, what can we do? Let’s just experiment and see. Do we like this? Does this look cool? Should we show it to some people? Yeah, people seem to think that’s cool, but it looks really – you can’t just have it coming through the wall through an animated reveal matter, because that’s not sophisticated enough.’ So then we were like ‘Okay, let’s develop this ripple, and then use this ripple as a displacement element to displace it.’ And all in 2D of course to displace the wall. So when they stick their head through the wall, it’s like the wall becomes viscous in some way, to try and make sense of the fact that they were jumping through walls and portals and stuff.
What we had were about 20 SGI VGX graphics systems in this tiny little space, it was literally a sweatshop, and we had to bring in a portable AC unit to cool it down because we had all these machines in there, and it got really incredibly hot. It was just a crazy, crazy time. Given a lot of these people hadn’t even used a computer before, but they were great rotoscope artists. I was teamed with Shannon Noble, and I would work during the day, and he worked during the night, and he would create all of the roto-mattes for me. For shots like, when she’s coming out of the wall, there had to be a reveal matte for that. So Shannon would paint all of that stuff, and he would do all the roto stuff. We’d work on the same machine, and we’d share the same assets and stuff.
It was crazy, what we were doing. Never been done before, and we had software that had never been used before. Not just that, you couldn’t playback anything real-time back then. Everything processed in the foreground, so once you got your composite together, you had to hit render and it would composite in the foreground. You couldn’t actually use the computer until it had rendered your version of the comp. I swear to god Shannon read War and Peace sitting under the desk on that show. We joke about it, but it’s actually not a joke.
We were working really hard, and it was really intense, but there was also a shitload of downtime which you don’t have today, in today’s visual effects industry, because the tools are basically in real-time, and even when you want to render something, and you send it off into the background to render, and you carry on working on something in the foreground. Well, of course you couldn’t do that back then, so you worked really hard, and really intensely, and then you’d have an hour while it rendered where you couldn’t do anything. You’d hang out, and you’d talk, and you’d go and see what everyone else was doing, and we’d help each other, and we’d share things that we learned.
All kinds of crazy things would happen – I remember I was trying to do a shot one time, and I was like, ‘I can’t do it. I need another light. I need to be able to put another CG light onto this to be able to get it to look how we want it to look.’ Gary Tregaskis would be there, and he’d write some code for us, and then suddenly I’d have another light available.
Every day was a new day. Every day brought something new that we didn’t know we could do. We were devising methodologies, and coming up with ideas, and creating stuff that we would just be like, ‘Oh, shit!’ And it was more like an experimentation, because there was no rulebook on how to do anything! Or how to achieve a particular look. So you would just experiment until you got something that kind of looked cool. Which, as an artist, as a painter or whatever, when you’re at art school … I don’t know if this holds true for everyone, but certainly for me. Sometimes when I paint something it just evolves. You don’t go, ‘Okay, I’m gonna paint a landscape and this is what it’s gonna be.’ You start painting something and it turns into something completely different. You paint over your painting because you don’t like it, and it doesn’t bear any resemblance to where it started, and it was all a bit like that with Flame.
After the film, I spent months with Discreet Logic showing them how we did things, and showing them what techniques we would use. It wasn’t like it is today, where you have a module that would do something, or you’d have a button that would do another thing. You literally had to have an idea of what it is you wanted to create, and then figure out if you could do it with the set of tools you had in front of you. It was a completely different world than it is today.
Afterwards, too, we managed to create this clique around Flame, like it was really cool. The way it was marketed with the T-shirts, and the buttons, and the badges, and the Logic magazine. It was just cool. And who didn’t want to be cool, right? It was an amazing tool. I believe it changed the world in our industry, in many ways, because it suddenly brought to the fingertips of anybody who hadn’t had a tool to do high-end compositing. Anybody could do it, if you could afford the system – which was ridiculously expensive – you could get into the business of doing high-end compositing.
Tales from the trenches
Sheena Duggal: You know, in the scheme of things, the only people who were really doing any kind of large scale digital visual effects work then were ILM on the feature side. So, for Super Mario, we were just like an independent group of people who had all got thrown together in this sort of sweatshop space and whacked out a movie, which was crazy. We built a pipeline to do it using software that had never been used before by hacking bits of code together. It was just mad when you think about it.
Chris Woods: Strangely enough, although it was certainly daunting and kind of scary I think for everybody at first because we had a real movie and a deadline. It wasn’t an experiment. It had to get done. It worked very well. I think mostly people had a good time. We all worked our asses off, but everybody got paid on the hour. I’ve never been a believer in people working all night and all of that kind of crazy stuff where they’re not compensated. It was a lot of fun, but then there was the whole drama of the film itself and dealing with all of that. There was no established procedure for any of it. I hope, and I think a lot of the people there enjoyed the fact that they were all part of that team of about 30 people that really took an early crack at what does it mean to do digital visual effects for film.
Kevin Bjorke: It was a very relaxed place. I remember sometimes I would bring my daughter, who was a toddler, to work. So she would hang out with everybody. I even brought my dog sometimes. We basically had a few small or a few large rooms where everybody would sit around in a circle and then we had two large rooms with everybody sitting around in a circle and I think there was a little office on the side for Chris to make phone calls in and an office on the side that was really I think sort of like a converted closet that our editor would work in. And then behind that we had a darker room where our matte painters would work because they needed controlled light to make sure that their colours matched, because they weren’t matte painting on computer. I mean that was like science fiction crazy stuff, they were just painting on a piece of glass but with oil or acrylic paints.
Sheena Duggal: One fun memory is that – Technicolor was doing our visual effects dailies for us, scanning the film back out to film, and then it would get processed – they accidentally sent us Jurassic Park visual effects dailies instead of Super Mario visual effects dailies. J.W Kompare, who was our visual effects editor, put it up on the KEM, on the tiny little screen on those KEMs, and I remember we were all crowding around just looking at this going ‘Oh my God!’ We’d never seen anything like it. Who had? We were all quite disheartened that day, as I recall, looking at how amazing that was compared to what we were doing. I think, in our minds, we knew it was ILM and all that, but we didn’t really appreciate how amazing what we were doing was in the scheme of things.
Kevin Bjorke: And we were like, ‘Holy buckets, oh my God!’ And it was kind of funny because we’d been talking a long time about the fact that we wanted to do a CG dinosaur in-shot – I mean, we had an animatronic one – and we were told no, no, no, no, no, no, no, no. And then they saw that movie and it was, ‘We have to get a CG dinosaur in the show,’ and we’re like, ‘We have two weeks until we’re shipping.’
Sheena Duggal: One final memory – we were housed in this building and Boss Film Studios was right behind us. One day we were sitting in this production meeting, and our supervisor’s going, ‘I can smell burning!’ We were all like, ‘Yeah, we can smell burning!’ And we all walked outside, and it was literally raining ash. The Cliffhanger set, which was directly behind us, had caught fire. It was quite dramatic as I recall. There was never a dull moment, I’ve got to say, on Super Mario.
Previously published on vfxblog.com
Illustration by Maddi Becke.Subscribe to the befores & afters weekly newsletter