Miniature ships, big impact

Behind ILM’s efforts to craft both the Razor Crest and Moff Gideon’s Light Cruiser miniatures for The Mandalorian.

The inclusion of real, physical ILM miniatures into the visual effects mix of Lucasfilm and Disney+’s The Mandalorian has given many Star Wars aficionados reason to celebrate, as well as the crew at ILM itself. Models and miniatures are, of course, a big part of ILM’s history, and there seems to have been a resurgence in the interest towards this ‘old-school’ technique in filmmaking in recent years.

But, like many VFX projects, getting those miniatures onto the screen—notably the Razor Crest and Moff Gideon’s Light Cruiser models—was a significant challenge for ILM. In season one, it would require a home-made motion control system alongside the Razor Crest model build. Then, for the Light Cruiser in season two, ILM had to work during a pandemic, and that meant having only a limited number of artists on the build working almost entirely from home.

To get a sense of what was involved in bringing these Mandalorian miniatures to life, befores & afters spoke to ILM creative director and visual effects supervisor John Knoll, who was responsible for the motion control and model mover rigs, and to season two’s miniatures team of John Goodson and Dan Patrascu.

Why are there miniatures in The Mandalorian, anyway?

Right now in the world of television and streaming shows, it is generally uncommon for miniatures to be used. That’s partly because they are considered a more expensive visual effect to budget for, and so remain mostly the domain of feature films (and are perhaps being used less so there, too).

However, during the planning stages of season one of The Mandalorian, creator Jon Favreau suggested to the ILM team that they could build a model of the Razor Crest to use initially for lighting reference, as a way of helping to base what was initially to be a CG model into reality as much as possible.

“I’ve had the experience of going out and doing miniatures and it can end up being quite expensive,” relates John Knoll, in terms of the move to build a practical Razor Crest model. “Stages have a lot of overhead. I was pretty sure that any effort along those lines of doing a full-on, traditional shoot was just going to get rejected as being too expensive. This was a tightly budgeted show, and the all CG version had already been approved. So I figured the only way we could afford to do this was if it was this total guerrilla-style, garage operation, where it was John Goodson building something out of his garage and me putting together whatever I could put together to do the shoot.”

And that’s precisely what happened. Knoll started his ‘guerrilla-style’ motion control build by taking an early CG model of the Razor Crest, and animatics of the kinds of shots that were planned for the show, to help him work out how big the model would need to be and how much track he might need to shoot it. The track, the camera head, and model mover were also designed and modeled virtually.

“I played around with a couple of different scales.” advises Knoll, “I thought if I built 50 feet of track—which is about the longest thing I can fit on our motion capture stage—and we had a 24-inch model, that would be the right balance. Then we could get shots that were dynamic enough that you get far enough away from it on the track and that we’d have a hope of getting enough detail in the model that would hold up for those closer views we needed to do.”

Knoll looked to 80/20, an aluminum extrusion based rapid prototyping system as the basis for for the track framework. He had also considered renting or purchasing an existing motion control set-up, but this proved too costly. So, in true Knoll fashion, he made it himself.

All of the build was handled in Knoll’s own machine shop in his garage, using CNC milling equipment the VFX supervisor had also developed to aid in the machining process. “I used to build motion control equipment and I wrote software to drive my own motion control system years ago,” he says. “So this wasn’t something that was alien to me. And a couple years ago, I did a hobby CNC project and that tech is all the same as motion control.”

The end result was a machined pan-tilt head and a three-axis model mover, as well as the necessary electronics and control systems. Interestingly, around half of the software stack Knoll relied on for this Mandalorian motion control system was old code he had written back in 1990 that ran on a ‘Tondreau’ system.

“Way back early on in my motion control career, I bought a Tondreau system—Bill Tondreau gave away a developer kit to let users write your own software that ran on this hardware. I put a few months’ worth of work into it and I still have that code. When I wrote this new system, I looked through that old code and I pulled the whole command line UI and move editing tools out of it and that’s what’s running on my current motion control system. So it’s a mixture of stuff I wrote just recently and vintage code from 30 years ago.”

The model mover gets (slightly) more complicated

Initially, the motion control system was not designed to do a full three-translation axis, ie. to enable a three-rotation axis on the camera or three-rotation axis on the model. Knoll says this was because “it was mechanically more than I wanted to build and would end up being too time-consuming and expensive to do. So I was really looking at, what’s the minimum capability that this needs to have to be able to execute the shots that the production’s VFX supervisor, Richard Bluff, and the filmmakers need?”

Knoll figured he could ‘get away’ with one-translation axis, where there would be the 50-foot track with just pan and tilt on the head and a simple two-axis model mover (like what was used on A New Hope) would be sufficient.

“That model mover back then involved having a blue pylon with a rotating mount post atop a turntable,” explains Knoll. “You’ll notice in the original Star Wars, there was never a shot where a ship was flying towards camera and then we pan with it as it flies away, because those shots inevitably want some roll in the ship. Unfortunately, there’s no way to mount the model where you aren’t going to see the mount. To do that roll, you would need to do a nose or a tail mount. If it’s flying towards you, then it has to be a tail mount but then if you pan with it as it flies away the mount is visible. If you mount from the off camera side, you’re not able to roll the model.”

It wasn’t until The Empire Strikes Back that ILM built a three-axis model mover—the kind that had big cradles that allowed the mounting of models from their far sides or bottom in order to realize shots coming towards camera and flying away. Knoll’s early thought was he wouldn’t be replicating that kind of thing and it would just be a constraint on the kinds of shots they could achieve with the Razor Crest model, that is, no shots could be done of the Razor Crest coming towards camera and then panning as it flies away. This constraint on shot design was part of the aesthetic on A New Hope and the intent was to embrace it.

“But then,” advises Knoll, “I saw their wish list of shots that they wanted to do and at least half of them were exactly that shot design! So I figured I had to bite the bullet and just have that third axis available. So, that’s where that model mover got more complicated and turned into a full three-axis thing with the cradle.”

Planning shots to stay inside the Star Wars world

Knoll had worked out the design of the motion control system in CG, both because of the obvious advantages of being able to rapidly try out different designs, bus also to help others (and himself) clearly visualize what it was that he was proposing to build. As things geared up, he wanted to take that a step further to ensure that the kinds of shots planned for the Razor Crest were the kinds of shots that reflected what could be done with motion control.

“I thought, what if we take that CG model, import it into Maya, rig it and attach a camera to the camera on the motion control system? If [production animation supervisor] Hal Hickel animates the motion control system, then by definition he’s not designing a shot that is beyond the limits of the system. And then we’ll be able to see if there’s a clearance problem where the camera’s going to strike the model or the model mover is going to intersect with the track.”

It was, says Knoll, all about imposing some physical limitations on spaceship maneuvers, just as the original visual effects practitioners on the optical days of Star Wars had placed on them.

“I do think a big part of the shot design aesthetic and look of the shots in the original Star Wars films came out of the technology that was used to produce them,” observes Knoll, who adds that one major difference between then and now is that artists used to really just start with the key element (a single X-wing model, for instance) and then build up a shot, rather than now where they have all the CG elements right there in a scene to use.

“The original motion control systems were recording devices. You didn’t set keyframes and interpolate the way you do in CG. You would record one axis at a time, like overdubbing in music. You’d look at the shot and say, all right, well the major axis is the track move so we’re going to start by programming that track first. You had a joystick with a potentiometer that controlled the speed of the motor, and you’d put it into record and look through the viewfinder while you dialed the speed of the motor, and record that axis. You’d put that in playback and you’d run back to the head, and then say, ‘Now I’m going to record the pan move.’ And you’d record that, and you’d keep doing it until you’d built up the whole thing. Those limitations, and that style of working, had a pretty visible impact on the designs of the shots.”

Another shot design aspect Knoll was keen to emulate from the original Star Wars trilogy was the sometimes seemingly incoherent—but still incredibly effective—relationship between foregrounds and backgrounds.

“Sometimes the motions have almost have nothing to do with each other,” states Knoll. “A lot of times, there’s an X-wing flying towards camera and there’s a star field that’s rotating and they’re completely different moves, and they’re really unrelated in that way. So we spent time studying that, and emulating that in The Mandalorian.”

The shots ILM had to do before ILM did the shots

With a budget and schedule made for season one’s Razor Crest shots, model maker John Goodson (who had started in the ILM Model Shop in the 1980s but since transitioned to a digital artist at ILM), had estimated he would need about four months to build the ship miniature. Ultimately, that build would come together via a unique collaboration of initial designs from the Lucasfilm Art Department headed by Doug Chiang, further digital design, readying of the digital model for 3D printing, final 3D printing, making hand-crafted and kit elements, painting and LED lighting installation. (There’s more on ILM’s model building process, below, where Goodson and Dan Patrascu discuss what went into the Light Cruiser for season 2).

While Goodson and others commenced their model build, Knoll was also delivering his motion control system in his garage. “I had just finished all the machine work so I had all the mechanics complete, and I was starting in on soldering up prototype electronics for it, when we got word that Jon Favreau had inquired about showing something at the Star Wars Celebration event in mid-April 2019 (season one of The Mandalorian would began airing in November that year). Our motion control shoot was originally planned for late May, so it really meant, could we have something a little more than a month earlier than we were all planning?

“My first call was to John Goodson,” remembers Knoll. “He said, ‘Well, what do you need to see?’ We had tentatively picked a shot that was mostly just seeing the front of the Razor Crest, and he looked at it and said, ‘Alright, I’ll just focus my attention on that part of the model. We don’t need to see engines, right?’”

Not having to see the engines would be a bonus since those required extensive detailing and electronics for the engine lights. While the mechanics of the motion control system were ready, Knoll realized the software and electronics for his system could not be ready for the Celebration shoot. “So I called General Lift’s Joe Lewis, and he rented me an old Kuper system from right out of 1992! It was great. It booted right up and it was just like being back in the 1990s.”

Having now established they could achieve the desired shots for Celebration ILM was buoyed by the opportunity to do some early shots. After giving it some more thought, Favreau decided he wanted to expand the offering, asking if ILM could do two shots, one flying towards camera and the next flying away. It meant, of course, that there was just about no part of the model that wouldn’t be seen. It seemed Goodson would have to ramp up the detail on the model, and quickly.

When a miniature is photographed, you are committing to a lighting scheme that’s not really possible to change later. The team wanted to see some lighting options and have an opportunity to weigh in on them before the day of the shoot and it was too late to adjust.

“John Goodson brought in the model to ILM as it was,” remembers Knoll. “I set up a still camera and started doing some lighting tests to figure out how I wanted to light the ship. At that point, he had clad the model in this aluminum foil and aluminum tape, and it was really shiny—it was super-glossy aluminum. Lighting to show form was difficult because you tended to get a black image with just a handful of overexposed highlights”

The solution was to dull the surface down to reduce the shininess and allow the highlights to spread and better show form, a task Goodson took on over a weekend, before returning to carry out the shoot for the Celebration shots. “Well, we shot it, showed those shots at Celebration, and it went over pretty well there,” attests Knoll. “In the meantime, John took the model back and did another month’s worth of detailed work on it. I sent the rental Kuper system back and finished building my electronics and got the software up to a workable state.

More links to the past

When it came time to shoot the actual shots of the completed Razor Crest miniature on the completed motion control set-up, several passes would be involved, as Knoll explains. “For key and fill, I would generally shoot those separately because we often wanted to rebalance them or take the fill and tint it a little bit more of a cool color. By shooting them separately, we had those controls. Then I would do running lights—there was a number of little red running lights on the ship, so we’d shoot those separately. Then, cockpit lights. And then I would do the engine pass as its own pass, because we would shoot those with a double fog filter.”

In fact, the use of a double fog filter came about from another linkage to ILM’s past. Knoll had recalled from his beginnings at ILM as a camera assistant that during filming of shots for Star Tours and Captain EO that engine passes on miniature ships employed double fog filters.

“I was looking at purchasing a set of those when I ran into Peter Daulton in the hallway at ILM, who had shot a lot of that stuff in his early days at ILM. I said, ‘I have a memory of you using the double fog two and three, does that square with how you remember this?’ And he said, ‘Let me look it up.’ The next day, he brought in a sheet, a cheat sheet from Return of the Jedi that had every model listed, along with what voltage you should use to power the lights, what filters were used, and what exposure was everybody’s favorite—because they wedged all these things out—‘Oh, this is good at one second at F16 and for this one, use the double fog two or the double fog three.’ So, the Millennium Falcon was shot this way, the B-wing was shot that way. From the list, it seemed like the double fog three was the one to use.”

The connection to ILM’s history didn’t stop there. The cameras used for the motion control shoot were Canon DSLRs, which Knoll was able to connect up to interface with his system (effectively setting up his micro-controller to do the equivalent of pushing a remote switch to fire off the exposure on the camera). While the camera bodies were standard, it was the camera lenses that took on an extra-special connection.

“During my research prior to our work on season one, I learned that the ‘cool kids’ were not using modern lenses on these things,” shares Knoll. “Part of the reason why is that the modern lenses, as a convenience, by default, when you’re looking through the viewfinder, have the aperture all the way open to give you a nice bright image. And if you’re going to shoot at a small aperture, they don’t stop down until you press the shutter button. They’ll snap down to the smaller aperture, take the picture, and then snap open again. That’s great for still photographers, but the problem is that the mechanism that does that doesn’t repeat.”

“I’d seen stories online from people who were doing stop-motion or doing time-lapse where they would see flicker in their footage,” continues Knoll. “If you’re wide open it doesn’t really matter, but if you’re shooting down at a small stop like I was going to have to do on these to maintain depth of field—if I was going to shoot at F16 or F22—I was going to experience that flickering problem because the aperture doesn’t repeat. What a lot of people were doing to solve that problem was using old lenses where the aperture is set manually.”

For Canon bodies, Knoll discovered that the preferred older lenses were NIKKOR primes because the mount is smaller than the Canon mount and an adapter ring can fit the body without changing the back focus distance. Knoll acquired the right adaptor, and all he needed now were NIKKOR primes.

“It turns out that [ILM senior VFX supervisor] Dennis Muren bought a set of those original NIKKOR Primes from the Kerner auction years earlier. He kindly loaned them to me for the shoot. Which means the lenses that I shot all of the Mandalorian motion control work for season one and season two were original ILM lenses, the same lenses as ILM’s Vista Cruiser camera system and on the original Dykstraflex. I loved that we shot these models with some of the original lenses from Star Wars.”

‘A return to the G-matte’, and other compositing fun

While many years ago the motion control passes of ILM’s models were composited in an optical printer, The Mandalorian miniature passes were of course composited digitally. An initial compositing challenge on season one for the Razor Crest miniature was that a stabilization pass had to be carried out due to bumps coming from the track joints (these were rectified for season two).

Also, garbage mattes became a thing, again, points out Knoll. “There were G-mattes that had to be made because there was all kinds of extraneous stuff in frame that had to be removed. It’s something you don’t normally get in computer graphics. So, it was a return to the G-matte!”

Other than these aspects, Knoll noted that the model work for the Razor Crest involved relatively conventional compositing. Ultimately, too, the majority of the Razor Crest elements in the show remained computer graphics, says Knoll. “On season one, there were 16 or so model shots, and on season two, we used a couple of elements from season one and I shot six new Razor Crest shots and then we did 24 of the Light Cruiser elements.”

Indeed, a significant effort at ILM went into ensuring that the CG model of the Razor Crest matched the miniature, and vice versa. “Even though the miniature is in a very limited number of shots” recalls Knoll, “those miniature shots made all of the CG shots better because we had tuned up the CG model to match everything we liked about the miniature. The way that John Goodson had scuffed up all the panels with some Scotch-Brite to get those nice, different levels of shiny and rough on it, and the anisotropic highlights on it were really, really nice.”

Last year, the team behind the model—from design to CG to practical—took home the VES Award in 2020 for Outstanding Model in a Photoreal or Animated Project. The award was presented to Doug Chiang, Landis Fields, Jay Machado, and John Goodson.

Perhaps in the same way that both a practical Grogu puppet and a CG Grogu have graced the screen in The Mandalorian without audiences knowing which is which, few have been able to tell the difference between the miniature Razor Crest and its digital twin. Even those who worked on the shots have struggled with this.

“I’m going to be honest, when I watch the show, I don’t always remember which shots are CG and which are miniature,” admits Knoll. “Some of them are really obvious where I’ll go, ‘Oh, I remember shooting that one.’ But some of the others are pretty generic fly-by shots and we did some of those with models and some of them with CG and sometimes it’s pretty hard for me to tell the difference.”

A new model: the Light Cruiser

Season two of The Mandalorian featured a new spaceship that was also both a CG model and a practical one—Moff Gideon’s Light Cruiser. To build it for real, ILM first had to work out how big the practical model would be, that is, what scale they would work at.

“I met with Doug Chiang,” recounts John Goodson, “and he had mentioned to me previously that maybe we would do a bigger ship for season two. I went in to look at the drawings and the renderings of the computer model they had, and then it was a discussion about how big do we want this thing to be?”

“They had a design for what the Light Cruiser would be and they’d built the 3D model of the basic design,” adds Knoll. “Doug Chiang had made two printouts, one three feet long and one that was four feet long. We laid them out on the floor and looked at them and started talking about it. I said, ‘I don’t think it’s big enough.’ I’d seen a couple of the shots that were on their wish list and saw we needed to get right on top of this ship and that we had to fill the frame with really small sections. So I thought the bigger we can go, the better.”

A budget was prepared for building a four-foot ship and a six-foot ship. “We compromised at the five-footer,” says Knoll. “I think that was big enough that we could get most of the shots we needed to, and wasn’t going to be so heavy that it was unwieldy on stage.”

“Once we decided on the size,” discusses Goodson, “then the art department printed out renderings for us that were orthographic. So we got a top view, bottom view, side view. Then Dan Patrascu and I started looking at it and we realized there were areas we couldn’t see. So we got them to print up some three-quarter renders of those. From that, Dan stepped in and began designing and building the armature.”

This armature of the Light Cruiser was engineered from honeycomb aluminum in order to try and keep the five foot model as light as possible. “We just started layering styrene and acrylic and some 3D printed parts on top of that,” outlines Patrascu. “There was a lot of fitting, assembly and disassembly, and assembly and disassembly, which was made a little bit more complicated by our ‘remote shops’, as it were.”

Those remote shops came about, of course, because of the advent of the COVID-19 pandemic in 2020. Crew members were now required to work from home due to lockdown restrictions, with initial plans to have several artists contribute to the build of the Light Cruiser revised to be primarily just Goodson and Patrascu.

“Dan would build parts like the engines, and I would build stuff ahead of it,” says Goodson. “I think once Dan started building the engines we only got to test fit them twice, and then I had them. So you’re kind of building blind.”

“It was a 37-mile drive between our houses, and there were quite a few fitting sessions,” notes Patrascu. “That was made even more complicated by, how do you work on a five-foot model when you have to stay six feet apart? That added an extra twist into things.”

How to build a Light Cruiser

To make up the necessary surfaces for the Light Cruiser, which resembles an Imperial Star Destroyer, Goodson began by relying on his own large inventory of kit parts. “The thing that’s going to be interesting is,” he suggests, “is that I’ve had people asking me, what are the parts? Well, I have no idea what they came from, because I have banker’s box just full of parts collected over the past 30-years!”

“The only thing I required here, though, was that I would go through and I’d find pairs, especially left and right. It was so I could have symmetry on the Light Cruiser. That was a critical thing for me, to find multiples of things, but I have no idea what this stuff came from. For the people who try and replicate this thing, sorry – good luck trying to do the kit parts…”

In addition, Patrascu designed parts and printed them on his 3D printer. “We stuck to traditional styrene and acrylic for major portions, but I picked out certain details,” he says. “Like the engines themselves, they’re fairly complicated. And I spent a good deal of time actually copying the artwork that we had, as close as I could, and then modeling those up in 2D space and printing them. I think I had my printer running for six months, 24/7, non-stop, just printing various odds and ends. Some things would take two or three days to print.”

The Light Cruiser’s bridge was scratch-built by Goodson, which proved to be a complicated part of the model build. “I left it hollow inside, because I didn’t know what we were going to do. In the past, we used to just put diffusion in the windows and they would glow. But not too long before we were going to shoot it, I was talking to John Knoll and he said, ‘I think we’re going to need an interior.’”

The bridge was designed to have wings on either side. Goodson made little doors that snapped into place about half an inch tall and an inch-and-a-quarter wide. “I had seen the set and I saw some photographs of it, so to replicate that I just took half-inch acrylic and used a Dremel tool and, just by hand, cut the shapes out that looked like consoles as blocky forms. I made a left and right bridge wing section and a center piece that was just one piece of acrylic with all these protrusions in it. Then I painted it silver and black, and then went back and scratched off little areas.”

“Each area,” outlines Goodson, “got an LED that plugged into the side of it. It was one of my favorite parts. It was really simple, but it looks really interesting. You don’t know what’s going on in the bridge, but it looks interesting in there.”

Another exquisite piece of detail for the Light Cruiser came in the form of tiny TIE fighters made for the ship’s launch bay. The TIE fighter fuselages were 3D printed by Patrascu, with the panels that hold the wings being machined on a CNC mill and fitted to the 3D printed parts before being painted. “They’re tiny,” notes Patrascu. “They’re very well-detailed, and you barely see them, but they are there.”

For the overall coloring of the Light Cruiser, Goodson stuck mostly to the ‘white’ look of a Star Destroyer. “I had primer-ed part of it, and Doug Chiang really liked that,” recounts Goodson. “So I sampled that primer, and had that mixed, and used that as a base coat. Then it was just airbrush detail, a little bit of pencil line, some classic Letraset detailing, for little, black tick marks. And then airbrushing to camera.”

There’s a rich tradition of model makers over the years adding hidden or secrets details to model surfaces, using unusual kit pieces or combining different ‘greeblies’ to form the final model. In the case of the Light Cruiser, time was not on ILM’s miniatures team’s side to do too many such things.

“At one point, I had a piece on there that was from Star Trek: The Motion Picture,” Goodson offers. “But it became detached, and it never made it back on.”

How to light a Light Cruiser

The Light Cruiser was integrated with nearly 300 LED lights, including for its engines. “That was a challenge,” observes Patrascu, “because there was so much power being used that my little power supply couldn’t keep them lit. I think the first time I powered up the engines, I sent a picture to John and his response was something along the lines of, ‘That looks really dangerous.’”

Lighting of the Light Cruiser’s three engines began with reference to the concept art. Here, there was a suggestion of a dot of light in the center of each engine. Goodson and Patrascu felt that there was an opportunity to scale up the effect by using the engine’s shape to form a unique pattern of light.

Recalls Goodson: “I put a light under the engine and sent a photograph to Doug Chiang and said, ‘What do you think?’ He touched base with John Knoll and they said, ‘Looks good, let’s do that.’ We went with a lot more light than what was initially spec-ed for that.”

Lights that served as points for the ships portholes and similar parts were built with fiber optics. “I did something that I’ve done in the past, but I’ve heard people say not to do it,” comments Goodson. “That was, I lens-ed the ends of the fiber optics. What happened was, there were a couple of times where we got a lens flare off a fiber optic that was hitting the camera. When that happened, I would just nip off the end of the little lens. Now if you look at some of the shots, you’ll see light cast out onto the hull. I think it brings a really interesting scale look, it’s not something I’ve seen before.”

How to shoot a Light Cruiser

The existing model mover made by John Knoll for the Razor Crest model—which enabled for three-axis movement—was something he determined would not be required for the Light Cruiser. This was because most of the required shots of that ship were static and would need only camera motion. Instead, a rigid mount was employed to position the model on.

The motion control system, on the other hand, was revised. As noted, the original tracks had small bumps on their joints which required post-stabilization work to be done. This time, Knoll looked for ways to mount the track segments with more precise alignment—“I machined some plates that held them very precisely so that there wouldn’t be those big bumps at the track joints.”

The main change to the motion control system was in its electronics. Knoll designed a printed circuit board, and then built that, as well as a ‘more professional-looking’ enclosure for the control system. “It was mostly to make sure that the whole thing was more reliable and there was less chance a wire was going to flex and break and take the system out of commission,” he notes.

Meanwhile, in addition to filming the Light Cruiser on stage against blue and black, a number of shots were actually achieved by shooting it outdoors on a rooftop, as Goodson describes.

“We got up on the roof, three stories up. And John Knoll was at ground level, with a telephoto lens, and photographed up, with the model against the actual sky. This was for where Boba Fett sees the ship and he goes, ‘Oh, the Empire is back.’ It was really effective.”

Miniatures are back

The miniatures and models experience on The Mandalorian has, clearly, been a positive one for ILM. Goodson and Patrascu, despite the unusual at-home working conditions on season two, certainly enjoyed the process. “I mean, we would text each other at 2:00 in the morning while working furiously on it saying, ‘Oh my goodness, we’re working on a model for a Star Wars show. This is really real,’” shares Patrascu.

“For me,” adds Goodson, “it goes back to when I started at ILM in 1988 and I was then in the Model Shop for 15 years. I switched over to computer graphics on Episode III. It’s really interesting to go back to what I was doing when I started at ILM because that was why I got into the business.”

“I’m hoping that we do more of this in the future,” comments Knoll. “It’s fun to do. It’s gotten a good response from the public. And I’m keen that this kind of way of working with models and motion control doesn’t disappear from the company. I’d like to make sure that some other people have been trained up on this and are familiar with how it all works so it can keep going into the future. It’s part of our heritage.”

Become a befores & afters Patreon for bonus VFX content