Viewpaint: ILM’s secret weapon on Jurassic Park
By Ian Failes
“All of a sudden, it’s close-ups! The T. rex head is in your face and filling the frame, and eating a raptor or fighting. So then we thought, ‘Oh, shit.’ – John Schlag, computer graphics artist, Jurassic Park
Most people will have heard about the Industrial Light & Magic animation test that convinced Steven Spielberg to adopt computer graphics for the full-motion dinosaurs in Jurassic Park. It was a seminal moment in visual effects history. But along the way, ILM made several other stunning tech breakthroughs for the 1993 film.
Among them was the continued development of a 3D texturing tool called Viewpaint. It allowed artists to create an image that was a snapshot of a texture-mapped model that could be exported to an image-editing program to paint or modify the texture. Then, Viewpaint would map the painted pixels of that still image back into the appropriate pixels in the texture map, and render the model. That’s how audiences got to see incredibly detailed dinosaur skin.
On the 25th anniversary of Jurassic Park, vfxblog goes back in time with several of the original creators and users of Viewpaint, a tool that would be used at ILM for many years, and for which its developers – John Schlag, Brian Knep, Zoran Kacic-Alesic and Tom Williams, received a Sci-Tech award in 1997.
The birth of Viewpaint
John Schlag (computer graphics artist): After Terminator 2 and before Jurassic Park, we were working on Death Becomes Her. We used Death Becomes Her to test a number of things that we were planning to do on Jurassic Park. I had written a 2D matchmove programme called MM2 that later evolved into Repo. We had to do a very painstaking matchmove of the body, the neck and the head to get each of those shots to fly. Anyway, I was working away on Death Becomes Her, and Tom Williams, who was then the manager of the new software department, he and I were chatting and he said, ‘You know, there was this cool paper at SIGGRAPH 1990. It was by Pat Hanrahan and Paul Haeberli (‘Direct WYSIWYG painting and texturing on 3D shapes,’ Computer Graphics, Volume 24, Number 4, August 1990). It was the first real demonstration of a 3D paint system. And Tom said, ‘Do you think we could do something like that in production?’
Tom Williams (supervisor of software and digital technology): There were two or three things I’d been thinking about before Viewpaint came to be. The first thing was a tool we developed internally during Terminator 2 called Make Sticky, which basically projected images through a sort of pre-rendered buffer onto polygons and patched vertices as a way of texture mapping the surfaces. It was great for the shots we had in T2, but obviously it wasn’t that interactive in editing. It worked particularly well if you had, like, a projection onto a surface that was from the angle you were projecting relatively planar, so you didn’t get too much in distortions. You could stretch things in the texture map that you were projecting, but it was kind of painful.
The other thing was, I’d had this experience with someone from ILM’s creature shop. Her name was Jean Bolte and she had been painting a tiger for a commercial. What she did was a combination of sort of sculpting and painting at the same time for this one project, which we didn’t get to in the original Viewpaint, but we got in subsequent versions. It was about being able to change some of the surfaces while you were doing the painting.
I was thinking, ‘Wow, Jean has to go through all sorts crazy things because there’s like this leg in the way and it’s occluding where she wants to carry this texture, and we don’t have to. Like, we can delete that leg temporarily, and just get rid of its visibility.’ So that was the kernel of the idea. Basically, the idea was, ‘Hey, we can render these pretty dense meshes and then be able to take whatever you’re painting and freeze it.’
John Schlag: So I spent a couple of months doing a prototype implementation. I just started doing some experiments, and I put together the first prototype implementation of Viewpaint. We had an in-house piece of software called Layerpaint at ILM, which I think was created by Doug Smythe. Layerpaint was a 2D or a two-and-a-half-D paint system that let you paint on one layer, and then have that pull paint from another layer. It was one of the very first programs to do this. Now, of course, that’s all built into Photoshop and it’s all old-hat, but Layerpaint was the first program I ever saw do this.
Layerpaint had been written there at ILM and we had the source code to it. I broke into Layerpaint and installed a shared memory interface so that Viewpaint could use Layerpaint as its 2D paint program, and then send paint information back and forth through this piece of shared memory. It was a very convincing proof of concept. Then we went to the purveyors of the commercial paint system we used at the time which ran on the SGIs – Matador by Parallax Software. And we showed them Viewpaint working with Layerpaint. I told them, ‘Layerpaint is not the paint program we want to use. We want to use Matador, but I had to use this shared memory trick to get the two programs to talk to each other.’ So I asked the Matador guys, ‘Could you please install the same sort of shared memory so that your paint buffer can be in memory that Viewpaint can read from?’ And they did.
Tom Williams: Parallax were very nice and let us do this because we didn’t want to go write our own paint system at the time. They gave us a bunch of back doors and gave us access to the buffers that they used when they were painting.
John Schlag: This was an era where most of the software at ILM would involve typing something into a text file, and you would run a command; it was not very interaction-friendly. Alias was one of the few pieces of interactive software that we had. And I had come from a background of MacroMind 3D and writing Mac interactive graphic software already. So when I got to ILM, I was kind of aghast, in some ways, at how primitive the interactive tools were. There were some standouts, like Alias, MORF, which Doug Smythe had written, and then Alex Seiden had written an interactive lighting tool. But really, Doug and Alex and I were the only people at that time bringing interactive software into ILM.
So when I put together Viewpaint, I put it together in such a way as to be interactive from the outset. Artist friendly. No command line, no scripting, no any of that. It was all about: bring in my critter, let me spin the camera around with the mouse, and then let me start painting. That was thing number one, and then just getting it to the point where we had a proof of concept. And I worked with the Matador folks to get the communication between the two programs going. And then my daughter was born, and I went on leave for a couple of weeks and Brian Knep took it over and proceeded to do great things with bullet-proofing it, changing how we did the mapping from screen space to texture space, that was great.
Brian Knep (computer graphics software developer): My role was to make Viewpaint more production ready. John did most of the work and then I took it over from him and worked with the artists to get it to be more like a tool that they wanted to use. For Death Becomes Her, there were some shots with Meryl Streep where her head gets turned around and so we had a 3D neck, basically, and the artists were using ViewPaint to paint on that. John Schlag had mostly gotten to a point where the artists were able to do that, but then it really got ramped up for Jurassic Park.
John Schlag: Previously, what you had to do to texture a 3D model was – and you can still do this in Maya today, only it’s a lot easier – what you would have to do was create what’s called a pelt. It was as if you’d skinned an animal and flattened out the skin, and then you could paint it on a table, flat. If you took all the bi-cubic patches that made up the T. rex, for example, and laid them all out on a table as squares of varying sizes, you’d have to paint on each one individually, and do lots of renders, and see how they would fit back together. It’s really hard to do a creature that way. Basically you would have to paint in a flattened texture space.
Brian Knep: The artists were spending a lot of time drawing in this kind of weird, warped way in 2D paint programs and then rendering it and then seeing how it looked. If it was off, they’d go back in again. It was just very complicated.
John Schlag: A really simple example of what used to happen is, take a cube and cut it so that you have that diagram, like you have a key of squares. Six squares, one on the centre, one on each side of it, and one extra one on one side. That’s a flat cube, and you can paint on that. But that’s pretty hard because you’ve got these edges that get folded back together, and you have to make sure that what you do is consistent along the edges. This was the original promise of Viewpaint, which we delivered on, which let you spin an object around in three dimensions and paint on it. It’s kind of like you’re painting on a piece of glass, and to this day, even MARI, that’s still basically how it works. It’s like you’re painting on a sheet of glass over a 3D render of your object, and then it slurps the paint off the glass and into the textures that make up the object. But it does that in the texture space of that object. You don’t have to worry about unfolding things by hand. And then you’re guaranteed to get the seams right.
John Schlag: When Spielberg came to ILM about Jurassic Park, we were just going to do long shots. It was going to be like the long shots with the Gallimimus stampede, and it would just be those small critters galloping in the background. And then one day – after we had done those secret tests – those shots started being, ‘Oh, it’s not just gonna be background. They’re gonna be foreground, galloping across frame.’ Then all of a sudden it’s close-ups! The T. rex head is in your face and filling the frame, and eating a raptor or fighting, or whatever. So then we thought, ‘Oh, shit.’
Zoran Kacic-Alesic (computer graphics software developer): I started at ILM in January of 1992, which was still pre-production for Jurassic Park. I was told Tippett Studio was doing the dinosaurs as stop-motion, but then we had been experimenting with the CG T-Rex to show Steven. So a bunch of my first efforts there were really on the enveloping and skinning for the T-Rex. Then all of a sudden we were doing these full-scale digital dinosaurs! One of the biggest issues was the amount of detail that you had to put in in order to make these dinosaurs look really good.
John Schlag: But that’s where Viewpaint came through for us. Because you’ve got a 3D render of your creature, you can paint more on it or push the button at any time, and what you can do is look at a render and say, ‘Oh, well just these patches up on the snout, or the schnoz of the T. rex, I need to repaint them at 2K resolution, whereas they were only 1K before.’ What you could do is just take the image file that represented a single patch. It’s just a square or a rectangle of 1K by 5-12, or 1K by 1K, and you just up-res the file and then render again. Now what you’ve got is basically the same information, it’s a patch which is too blurry because it’s too close to the camera. But now you can go in and re-paint that patch.
So having Viewpaint basically freed us. You can do this with any creature and any texture space, once you address the seam issues. Viewpaint made it easy to go in and just say, ‘I want to up-res this patch and re-paint it at 2K or 4K.’ So that’s how we got from long shots to medium shots to close-ups.
Zoran Kacic-Alesic: At one point, John Schlag actually switched from software production to doing shot production work, so I jumped in to help take Viewpaint from concept through to a usable tool. Most of my contribution was taking what John and Brian had started with and basically getting it to work for artists. There was a lot of improving the algorithms, getting the UV mapping to work. I was also splitting my time between work on enveloping and skinning for the dinosaurs.
Tom Williams: It was not a fast system at first because it would take this pretty dense model that we were using. And in fact, it was so dense that for the surfaces for Jurassic we had low-res versions and blocked out versions that we used to do all the animation, so you could sort of real time get feedback on what you were animating.
Zoran Kacic-Alesic: One of the first challenges was the texture layout. There were really no tools for laying out textures, so we were basically using the same tools to layout geometry in the rectangular periods for enveloping. And then that same information was also used to do the stitching between the surfaces, and then the texture layout was basically using the same organisation of base line surface approach.
John Schlag: The animation was being done in Softimage by that point. Softimage had proven itself to the point where we went ‘whole-hog’ with it for Jurassic Park, and then what we had to do was write an inordinate amount of basically glue, holding together Softimage and Alias, which I think was still being used for modelling, and our in house pipeline. We were getting textures done in Viewpaint, and then things were being rendered in RenderMan. But there was just a prodigious amount of glue ware necessary to keep all these things talking to each other.
Zoran Kacic-Alesic: In some ways Viewpaint wasn’t a true 3D paint, you couldn’t really paint in a 3D view. We just didn’t have the power to do that at the time, so what was really happening was we would bring in the model in 3D and then paint in a particular 2D view. So you could spin your model, do a lock, and then freely paint in 2D. So really it’s 2D on a 3D view. It was really tough to propagate pixels from a 2D projection, back into the 3D map. There were also filtering issues there, and I remember a certain issue with cracks during rendering between the surfaces. So just getting the pixels to map across these separate surfaces that were stitched together to get it in a render – that was a little bit of a challenge.
Tom Williams: The last time I saw Viewpaint, it was pretty interactive. It was rotate and paint, and rotate and paint. Later on I went to Alias, and we did their 3D paint system there, which was based on a different technique just because the hardware would support it at the time. And we could also do things that I wanted to do at ILM that we couldn’t do.
John Schlag: The other thing I’ll mention about the very early history of Viewpaint, besides getting the architecture going and making sure it was interactive and artist friendly – the other thing I did for Viewpaint was I named it. I thought, ‘We’re gonna need a good name for this, and that will help market the thing.’ So coming up with ‘Viewpaint’ I thought was kind of a minor master stroke, because you can still see – not in today’s films, but for ten, 15 years – you could actually see credits for Viewpaint artists in films.
Getting Viewpaint into the hands of artists
John Schlag: I started working with Carolyn Rendu – she was our ‘user zero’ of Viewpaint. I believe the first creature that got Viewpainted was the raptor. It allowed her to sketch over the whole body first, and then also paint in flattened texture space, just to get a take on where the basic elements of the paint anatomy should be.
Tom Williams: Even before Carolyn, we had the art director, TyRuben Ellingson, come paint one of the dinosaurs as a test. And he knocked it out in like a day. And it was so far beyond what we were capable of doing. He would come in in his overalls, and he was not exactly used to the darkened rooms that we were all working in. He’s used to the art department where everything was light. So, we got him to come into the dungeons and try it out.
TyRuben Ellingson (visual effects art director): The existence of Viewpaint came into my awareness as part of a larger conversation regarding Jurassic Park production needs. At that point in time, the role of the visual effects art director was primarily concerned with VFX storyboarding, creating keyframes and/or concept art, and participating in daily shot reviews. However, as CGI evolved inside of ILM, new kinds of opportunities opened up, and as such, it was logical to take a fresh look at responsibilities, roles, and how things got done.
Before J Park and the breakthroughs that came with it, if asked to assist in developing skin colour patterns for a dinosaur, as VFX art director, I would have first headed off to the art department to generate a series of colour marker studies. Once finished and approved, that artwork would have been passed along to someone working in the model shop who would then paint the texture onto an actual physical sculpture of the dinosaur (non-digital). After that, the painted sculpture would have been photographed and the resulting prints shared with others involved in the creation of various effects shots.
All this changed with CGI, however, as it allowed an art director to work directly inside of the digital pipeline, and in so doing, ‘skip’ steps that were the normal best practices of VFX production. For Jurassic, Viewpaint was the tool that allowed me to enter into the evolving digital production stream to paint skin patterns directly onto 3D dinosaur assets – which could then be used in shots and approved as an element (of the larger shot).
Carolyn Rendu (digital artist): I had been at ILM doing traditional optical rotoscoping, and then I’d started doing working in the computer graphics department. I wandered through and saw that first test they’d done with the T-Rex that Ty Ellingson had painted and they threw it into a background plate. And it was pretty impressive.
TyRuben Ellingson: For that early test shot, which was of several Gallimimus running away from a lumbering T-Rex, I recall being provided a side elevation view of the Rex as a Photoshop file, to which I simply added the skin colour (using Photoshop). I think that file was set up and given to me by Stefen Fangmeier, or it could have also been Mark Dippe. Clearly, what I painted was used as a projection map of some kind, however, as there were a lot going on in that test and in those early days, I was very new to the technology, it was all kind of mysterious.
My main memory is of working very fast. I recall making use of a computer in the effects animation department — not a place I normally would have been working — and painting the map over the better part of a day. Next thing I know, it’s in the shot. The speed at which things happened in those days was really something new and very exciting.
Carolyn Rendu: How I came on board Jurassic was, Death Becomes Her had ended and one of the compositors, Steve Price, who has since sadly passed away, he happened to mention to me, ‘Oh, you have a painting background. They need someone to paint dinosaur skin.’ My background was in botanical illustration. I had also done some work with palaeontologists, and I love dinosaurs, so I said, ‘Why, sure, I’ll go.’
I went to [visual effects supervisor] Dennis Muren and he said, ‘Okay, well, go up and check out the software.’ And John Schlag showed me this Viewpaint software, and I just knew right then. I said, ‘I want to do that.’ I was lucky that I was there, because to happen to be there, to know the company, know everybody, know how production works, and have this technical drawing background where I was used to looking fine detail in nature.
Brian Knep: Viewpaint was kind of intuitive, which is why I think people liked it. It let you grab a 3D model of, say a dinosaur, that you could spin around. You would render it using the texture but we would also render a couple other passes. So what we would do is we would basically render it as a UV map. So it was an image of the dinosaur but every pixel, instead of it being the colour of a flesh or whatever or fur or anything, it was the UV map and a texture identifier. And that would go to a paint program – Matador – and people would paint on that. We would suck that back in and we would know for each pixel where it mapped to on the original texture.
Carolyn Rendu: My memory of first using ViewPaint was that it had Matador as the UI for the painting. Then there was this back-end into the 3D modeller software. I could move my model in any way I wanted to see what angle I was painting. Then I would call it into Viewpaint, and then I’d paint on that projection, which was still for a long time the classic way to do it. Once you projected it, and it was in your texture map, then you would move it. But what was really nice, too, about Matador is then you could also work on the texture maps separately.
Scores of painters? Not really
Carolyn Rendu: Originally it was me and Steve Price and Ty who would be working on the 3D texture maps. But Steve decided to do animation, and Ty was a bit like, ‘I don’t know about this.’ The process was very tedious, believe me. You needed a certain mentality to do this.
TyRuben Ellingson: What I do recall as being a little bit annoying was the amount of time it took in Viewpaint to save your work. Once you typed in a save command, things would just sit there for what seemed like a couple of minutes, but in point of fact, wasn’t likely that long, it only felt like it. Saving your work often was very important, as Viewpaint would crash from time to time.
These things as they were, it shouldn’t come as a surprise that crashing software wasn’t the only culprit when it came to losing work. From time to time, the power would simply go out, especially in stormy weather. I recall Janet Healy, the show’s VFX Producer, making her way through the halls and calling out, ‘Save your work, thunderstorm!’ It was all great.
Carolyn Rendu: It didn’t have the bells and whistles that Photoshop had, but Viewpaint was more robust. And it worked on an SGI. The simplicity of it was actually a benefit. It did everything that you really needed it to do to get the job done even though it might not have had the bells and whistles of Photoshop. You’re under the pressure of production, and you just want a workhorse, and Viewpaint was a workhorse comparatively.
TyRuben Ellingson: As Viewpaint only ran on an SGI machine, I was provided a workstation in the computer graphics department on the opposite side of the Kerner Campus from the art department, which I thought was pretty great as that’s where all the action was. Unlike the model shop, optical department, or stages, the computer graphics ‘department’ wasn’t really a specific space or place, it was more a ‘zone’ inside of a two story building which also contained the large screening room, corporate offices, machine room, and infamous ‘Pit’. The entire operation had a kind of ‘make shift’ feeling to it. When needs arose, additional space would be found, work stations set up, and cables run through walls or into the ceiling to connect everyone together. To me, everything felt fluid and organic.
Carolyn Rendu: I have to say this, I’ve never ever worked on a show that was so well prepared that was so thoroughly researched. They knew they were going into unknown territory. Every single person that I dealt with on that show they brought 110% of themselves to that show because we did not know how we were going to do this.
‘Here, paint this’
Carolyn Rendu: For textures, we mostly looked directly at what Stan Winston had built with the life-size dinosaurs. And I had a nice big Wacom tablet and a pen, so I just sat there all day long watching the screen and working away texture painting, doing mostly the T-Rex and the raptors. And, boy, it was tedious. You’re painting a little part, and then you stick it on there, and then you have to look and see. And then you maybe correct it, and then you’d turn the model, and then you find something. Then you paint a little bit, and then you have to stick it on. Then you move the model a little bit.
Actually, in relation to where we got the textures from, years later I was talking to one of the producers from Amblin, and she said to me, ‘You know, the sequence they were worried about the most was the kitchen with the raptors because they were cutting back and forth between the practical suits and the CG model.’ But then she said, ‘Oh, my God, when we saw that it was actually going to work we were just so relieved.’
So, just to do those raptors, they actually brought that full-scale raptor model – it was 13 feet – and put it into my office and said, ‘Here, paint this.’ And I’m looking at a wire frame or a white marble skinned raptor and it just seemed like an impossible job. But what I did was, I had a little Post-It that I made into a little pointer on the head, and I go, ‘Okay, I’m right here, on this scale.’
Above: the Stan Winston raptor that Rendu referenced during her texture painting, as photographed in a storage room in ILM’s C building. Photographs courtesy TyRuben Ellingson.
And then I would go to the model and find where that was and then I would mark that and I would move that around on the head because I had to make that head look like it matched the actual model. The colouring had to be just as precise on the face because you see those faces really close up. So I just worked my way inch by inch over that 13 foot model in the computer.
Crack! A near-disaster
Carolyn Rendu: As I said, the painting work was very tedious. And Ty was more than happy, I think, for me to do a lot of it. But then one day, there was an accident while they were all filming some reference for the Gallimimus running away from the T-Rex.
TyRuben Ellingson: Well, in those early days of Jurassic Park production, Phil Tippett, who is credited as dinosaur supervisor, but in my mind really functioned as director of animation, was very concerned that CG animators didn’t understand physical motion as well as stop motion animators. Because of this concern, a good deal of effort was put into creating opportunities for the Jurassic Park production team to get up from our desks and ‘strengthen’ our physical awareness.
To this end, classes in pantomime were held in a empty storage bays in which we did exercises that required we move in unusual ways. We also did a lot of play acting, adopting the behaviour of various animals encountering one another in a peaceful passive manner or taking on the more dynamic role of predator or prey.
When we started working on the gallimimus stampede which culminates with frantic dinosaurs leaping over a giant fallen tree, at some point along the way, the idea bubbled up that we could learn a great deal by acting out the sequence and filming it. Using our newly acquired ability to ‘get into the animal mindset’ and some very large drain pipes propped up in a configuration very similar to the fallen tree, the cameras rolled and we ran as a ‘herd’ of scared reptiles. We were jockeying this way and that, back and forth, and then up over the big plastic obstacle.
As a kid, I spent the bulk of my time working in my father’s basement studio drawing and painting, and as such, not too much developing excellent athletic skills. Jump ahead 20 years to that day out on a back ILM parking lot, and it’s not all together surprising that this gallimimus couldn’t muster the required prowess to escape the imaginary super predator closing in on me. When the moment arrived for me to make my leap, I caught a toe on the plastic log and quickly found myself heading face first towards the black top. To stop what I have to imagine could have been a much more damaging fall, I thrust out my right arm – yes, my drawing hand – and somehow used it to break my fall. Once I’d rolled to a complete stop on the ground, I knew there was a problem and shortly their after found myself at an emergency room.
The fall had shattered the elbow end of my radius, and as such, I needed surgery to remove a couple inches of the bone. Since that day, I’ve only had one bone that connects my hand to my arm — which is pretty strange when I think about it. The positive that came out of this otherwise sad little story, is that, upon seeing the footage of me taking my fall, Phil Tippett declared, ‘If Ty took this unexpected fall, then one of our Gallimimus’ would most likely have done the same!’ So indeed, in that sequence, if one looks closely, there is a Gallimimus who takes a tumble, however, rather then ending up in the hospital, it leaps right up and continues out of frame.
Carolyn Rendu: We worked in the same office for a while, and believe me, Ty was not sorry to not have to do this. I can tell you that, he was not sorry. He was perfectly happy to go off and do something else, and with this little bandaged arm. I still think he’s one of the most talented people I’ve worked with.
TyRuben Ellingson: All things considered, when one considers the magnitude of Jurassic Park’s success and impact on the history of film making, this story isn’t a bad one to have as part of my history.
The blood memo
Carolyn Rendu: I’d been working away on the texture painting and one of the things I had to paint were blood maps. This was where the dinosaur’s skin and flesh would be torn away. One scene where that was going to happen was in the rotunda where the raptors attack the T-Rex. Phil Tippett was really involved in what this would look like, and he knew what audiences liked, and so he had all these ideas about streaming blood, and blood here, and blood squirting.
Now, one of the things was, they had to work out, from shot to shot, the idea of blood continuity. So Phil wrote this memo to Steven Spielberg, and it had things like, ‘Blood in the mouth, blood on the dead raptor, blood on the T-Rex from raptor gouges’. One of the notes even says, ‘There will be no entrails, explosive geysers or buckets of blood flung about as a result of violent T-Rex shaking.’ And then it finishes with, ‘Please let us know if the blood plan meets with your approval.’
Well, this memo came back from production and Steven which was basically saying, ‘Not much blood.’ I kept all those memos and the original storyboards and just thought they were pretty hilarious.
Rise of the ‘Viewpainter’
[Jean Bolte is texture artist and supervisor at ILM, and regularly received ‘Viewpainter’ credits on the feature films she worked on. But she actually started in ILM’s model shop.]
Jean Bolte: My first memory of seeing Viewpaint was looking at an early T-Rex test on the screen. I was working full time in the model shop but was checking out the work going on in the computer graphics department, which was only a few dozen people at the time. Spaz (Steve Williams) asked me if I would like to see a test shot. I sat there in ILM’s C Theater with him and few others and witnessed one of the first times that anyone saw the Jurassic Park T-Rex lunge out of some bushes and head toward the camera. It was rough, but astonishing. I had seen the work on The Abyss and Terminator 2, but this was different. Having a creature that was painted: aged, complex, with so much detail and life in it had changed the game. That day I inquired how I could train to be part of the CG department. First choice of discipline: Viewpaint.
My background in animatronics and miniatures helped me because I was used to sourcing real things to do my job. I had learned a lesson in the model shop: don’t rely solely on imagination or memory – It’s invariably limiting.
I still make it a point to find actual objects or study animals and people and photograph them to see how they look, how they behave. I don’t just google something and make do. It’s important to get out and see things in situ: go look at the underside of a bridge, find minerals, bones, study a bird’s wing. You’ll discover details that will help you to feel like you can almost smell the thing you’re working on. That’s when you’re getting somewhere.
When I started working in CG it took some mental adjustment when I realised that one needed to tell a computer everything, nothing came for free except things like mirroring, duplicating and versioning. But the essential ingredients of a surface needed 100% input: there was no wood grain to reveal, no way to squirt alcohol on a layer of dirt and step back and watch the results. More and more we have tools that allow us to recreate gravity, expose layers, etc, but in the beginning it felt like the computers were in their own intimidating way somewhat stupid.
My favourite character to work on was Mulgarath from Spiderwick Chronicles. Not a widely known creature, but a pleasure to work on. He came to the screen with very little alteration from his initial design by the great creature designer, Carlos Huante. Carlos drew him so that that no matter what pose Mulgarath took he would form an X in his profile, which gave him strength and integrity as an evil character. From the sketches I knew he would need a number of natural elements and I gathered up drift wood, animal horns, manzanita and snake skins and photographed them for the paint. I even incorporated photos of my own fingers. These photos were shared by Carlos and the visual effects art director, Christian Alzmann, who used them in the subsequent designs, so we were all in synch on his look, almost from the very beginning.
Today, MARI is our primary paint tool but we have incorporated something we call MARI lookdev which allows us to see the paint applied and shaded (in real time) much the same way it will look when rendered. This is a huge step forward, particularly for viewing the specular effects of our surfaces but in addition we can see our layers of grunge and bump as well as other effects. We also use Photoshop, ZBrush, and Katana for lighting and rendering. We have an in-house software platform called Zeno, which one of our software designers, Collette Mullenhoff (who in 2017 won a Sci-Tech Award) is upgrading continually to give us best possible UV layout tool.
The impact of Viewpaint
John Schlag: Viewpaint had a good run. It was actually longer than I thought it would be, and when MARI came along they finally took the trouble to write a full-fledged paint program within a 3D paint program. And that was the killer combination.
Carolyn Rendu: Once I had left ILM and was working down in L.A. I saw what other studios were using and different styles of paint programs, and none of them came close. One of the first big issues of commercial programs was that they didn’t really understand production. So the value of having that created in-house was that they knew what a production tool should be and how you had to rely on it, and if you couldn’t rely on it, it was crap and no good, and nobody was interested in that. This had to work. That’s the wonderful thing about working at ILM. All anyone cared about is, ‘Does it work?’ And we all had this goal, and we worked as a team, but they understood production and the needs of productions.
John Schlag: I definitely had the feeling at the time that we were doing something big. It continually blew me away that things that were 13 inches wide on a monitor in my office were going to end up at 40 feet wide on theatre screens all over the world. We worked very, very hard, many times evenings and weekends. And things were still going to crap at that point, and that was a very unforgiving process. But once it came out, peoples’ eyes just lit up when they saw the film. Because the book is about bringing dinosaurs back to life, and we made that happen. It’s a movie you could go see with family and friends, and there was almost nobody who didn’t like this film.
Brian Knep: Winning the Sci-Tech awards in 1997 [Knep also received a Sci-Tech prize that year for his part in the Dinosaur Input Device] was a huge honour. It happened after I’d left ILM and it was a really nice, for me, a nice feather in my cap in a way, on that whole period in my life.
Zoran Kacic-Alesic: Helen Hunt was hosting the Awards and she nailed my last name absolutely. I was so impressed because most people have a hard time saying my last name, and so somebody called me one day from the Academy and asked me how to pronounce my last name and that was it. And just absolutely nailed it completely, I was like, wow, I gotta shake her hand and say to her, ‘Thank you. You really did a good job.’
Carolyn Rendu: Some of my last couple of shows at ILM were Jumanji, where I was working on the CG lion, but I got taken off that, and then DragonHeart, but then I had to leave. I remember I was just so brokenhearted that I had to leave my models behind. Inside the mind of a Viewpainter is kind of weird.
Illustration by Aidan Roberts.
More from vfxblog’s ‘Jurassic Week’:
The oral history of the Dinosaur Input Device or: how to survive the near death of stop-motion
The surprising game-changing VFX of Jurassic Park III
Back to vfxblog.com