Posted on Wednesday, January 18th, 2017 by Peter Sciretta
Last week I had the opportunity to jump on the phone with the visual effects team behind Rogue One: A Star Wars Story. After the jump, you can read a transcript of my full interview with Rogue One executive producer / visual effects supervisor / writer John Knoll, animation supervisor Hal Hickel, and special effects supervisor Neil Corbould.
In the conversation, Knoll talks about his original low-budget pitch for Rogue One. We also find out just how many people and how much time it took to recreate Tarkin, just how much work was put into resurrected Gold and Red Leader from A New Hope, and if we’ll end up seeing more deleted scenes with finished visual effects on the home video release, and talk Princess Leia, miniatures, and much more.
Rogue One VFX Interview
Peter Sciretta: John, last time I saw you I was sitting next to you on the set of Jurassic World. I was geeking out with you about visual effects and new Star Wars movies, but at the time I had no idea that you were developing the first Star Wars standalone movie. Before we get to the VFX, I’m interested to hear about your initial pitch. The Art Of book mentions it was a much smaller, lower budget story. Could you elaborate on how it was different?
John Knoll: Yeah, well in the original way that Kathy [Kennedy] was describing these standalone stories, she was thinking of them as smaller, scrappier productions with potentially somewhat lower budgets. That could afford to be a little more daring in experiment, tone, and genre. And so in the original conception of this, I was figuring well we can’t afford to go to that many planets or to take a lot of vast, expensive things to produce. So I pitched it more as a spy thriller than a mission war movie kind of thing. And gradually over the course of the development of the project, it sort of expanded and expanded. Disney sort of made it clear that no, that we wanna do amazing stories, go for the best things that you could make. You know, and the budget will be there for it.
Everybody’s talking about how you guys brought Tarkin back with CG A lot of people think of what you guys do in CG as “oh it’s done in computers.” Computers make all this happen. But the reality is people make all this happen. Right? So how many people worked to create that Tarkin that we see on screen?
John: How big was the crew? Like 16 people with all disciplines included?
Hal Hickel: Yeah. I mean, for the number of shots. For the difficulty of the work it essentially was a very small creative…. But just really great, talented folks.
And how long did that crew work on those shots?
John: Well, we knew that we had this character in the movie probably six months before principal photography began. So we started working on building the assets really early on. And the hope was that we could start developing some first concepts early enough that everybody would launch to post with confidence that we could pull it off. But it’s a very time-consuming and challenging problem. So it was in development for a very long time. And we were working out animation rigs and how to drive it well pretty late into the game. We started on it in January 2015 or so, and then we shot the scenes, and they edited, then locked down, and we got the scenes turned over. I think the first ones happened in spring of 2016. And then it was flat out from there.
Hal: But it this way, we had someone on the crew who was pregnant, and she created an actual human being in less time than it took us to create a digital one.
Was there ever any thought of having Carrie Fisher reprise her role as Princess Leia on set and then digitally de-aging her like what had been done with like Robert Downey Jr. in Civil War?
John: No. Unfortunately, there’s a time window, a time span within which those kinds of techniques can work and when enough time has elapsed you’re really not the same person anymore. Your voice isn’t the same and the way you walk and carry yourself is really not the same. So we felt like the most successful approach to simulate, to recreate an 18-year-old Carrie Fisher was to have an actress that’s closer to the correct age.
Was ever any search to find an outtake of Carrie Fisher that could have fit? Or was it always the plan we’re gonna do CG Leia?
John: No, we always knew we needed to create one. You know, part of the research doing the character, I’ve seen all of the dailies of everything of Carrie Fisher from New Hope and there really wasn’t anything we felt we could use.
Fans were very excited to see Red and Gold Leader return. What’s the process of scanning and cleaning the old footage to make it look like it fits in a movie from today?
John: Yeah, well that certainly was more challenging than it might appear. Because this film was shot on Alexa 65 and the latest generation digital camera with 17 stops of dynamic range and a really good signal to noise ratio that’s almost grain free. And to take this 40-year-old footage that was a bit grainy, it was kind of underexposed. The negative had faded somewhat. To try and get that to match in without really jumping in terms of its image quality was a significant compositing challenge anyway.
Now that first shot you see of Red Leader was a particularly hard one because it had been underexposed and the whole side of his face was completely black. There was no detail in there. And the other ones were not as bad, so we were in much better shape. And so what we did is we took one of the other setups of Red Leader where we did have shadow detail side of his face and we kind of cut and pasted in sort of Frankensteined together a frame of Red Leader that had the sort of appropriate detail. And motion slowed that in and that’s what repaired that side of his face. It was all de-grained, and there was a lot of color corrections done through mattes to compensate for negative space. And then lastly the same underexposed and grainy problem existed for the rest of the cockpit. And so we sort of decided to cut the losses on that. We rotoscoped around them, and the whole cockpit interior is a render C.G. model.