Zoic Studios on Building a VFX Environment Without Overwhelming the Clothes
JOHN SHIRLEY: We were brought on at inception. We were working with Joel Pront, who is a director from Australia represented by Believe Media. We sat with him and discussed different ideas and different concepts of pop-up books and the physics of a pop-up book. At the end of the project, we probably had 40 pop-up books sitting in the Flame bay. Once we got shooting boards from Joel, we set out and did pre-vis in Maya for a week.
How detailed was the pre-vis?
It was just a layout that gave us the basic camera movements. The people and the animation weren’t rigged into the pre-vis. It was like the shooting boards with camera moves in them. And we prototyped the book-close at the end. Joel actually has a design and effects company called Collider in Australia. He’s very savvy about effects, and we really like working with people like that. The shoot was very true to the pre-vis because we spent a lot of time in Maya working out how he wanted to see it. A lot of the times we’ll get a storyboard and bid it and do all the set-up for it and then you get to the day and things change, but Joel was very good about sticking to it. I brought my laptop out there, we watched each pre-vis as we set up, and we executed it. It was very easy to shoot.
And the shoot was just the models on green screen?
On one of the spots, the guy for Structure clothing closes a car door. So on the green screen floor we’d put tape in the dimensions of a car, and we’d set up C-stands or some kind of objects for him to interact with. While they were adjusting the lighting for the next set-up, I’d run out there and adjust that stuff. We shot all the green screens – basically four :15s – in one day.
The cinematographer was Dion Beebe. He’s pretty great. Was there thought given to the kind of lighting on the models, and how that would relate to the final look?
Absolutely. At the end of the day, it’s a clothing commercial. I can get wrapped up in the VFX, but Dion obviously has won an Academy Award, and he knows how to light properly. That helps, because clients can get a little jumpy about shooting on a green screen. It’s hard for them to see things in context. With him on set they felt confident that he would get what they wanted out of their clothes, and I knew he would get what I needed from the green screen. We set up four or five lights on the green screen and basically left those alone, and then he could light the clothing and his models the way he wanted to. We were lucky enough to have a good 15 or 20 feet behind the actors. We didn’t get a lot of spill cast or a lot of contamination from the screen, and he could concentrate on it being a pretty shoot for his models.
Where did the still photographs for the pop-up elements come from?
Once we finished with the pre-vis, we had a layout for what we wanted in the scene. For one scene, we’d say, “OK, we need a Jeep, a fire hydrant, a bank building, a hot-dog cart, and a street.” We worked with the production company and set out a task list for a still photographer. I think they had one in New York and one in L.A. going out on element shoots for us. We had to get the shots orthogonally. If you were to shoot the doors of a car, you’d shoot at 90 degrees from the plane of the image you’re shooting. We shot probably 30, maybe 40 DVDs worth of still images we had to go through – thousands of images. Even on set we were getting DVDs. We would print thumbnail images of the contents and, while we were breaking down for the next shot, hand them to Joel. He and his art director would approve what they wanted, and then we’d come back to Zoic and convert those selected raw files to TIFFs. Once that happened, those selects were taken into Maya and cut out, with a rough, almost scissor-like cut around the image itself, and extruded a couple of pixels deep so it had a paper-like feel, with some depth to it. It was a huge data-wrangling job.
No matter what, something always comes up. “We need a street sign.” So our lead 3D artist would have to run out at lunch and grab a picture, come back, convert it, cut it out in Maya, and put it in the scene. It was super-collaborative. The whole FX team was taking part.
Were any elements created from scratch?
I don’t think there was any image that was painted as a texture. They were all photographs. I know we did a lot of Photoshop retouching. On top of all the modeling and data wrangling we did for all these images, we had to be very cautious. We had to make sure that, when we shot a hot-dog stand, all the logos were out. We had to make it generic. That was pretty intense. But it worked out.
What was the compositing workflow like?
Two of the spots were done in the Flame, and a couple of them were pre-composited in After Effects and Combustion before going to the Flame to have the final look applied. We would have an AE or Combustion compositor pre-comp the whole scene. They would pre-composite the backgrounds together and make sure they’re looking good, and then basically comp the person in the scene and get that looking good with spill corrections and color corrections, and then render out those separately. I’d get a composite with a matte for the foreground person, a matte for the buildings, a depth-pass for the CG, and then I’d apply the look of the clients in the Flame.
The client wanted the clothing to pop from the background. They wanted to showcase it. But we didn’t want it to look like it was mashed together, so we were cautious. We had to make their clothing stand out from the book itself, but still have it look like it was there in real life. So we worked with Joel on different sharpening filters and treatments to the book so that it got the texture of paper. We had a macro set up in Maya that took a little bit of the color out and did a little bit of filtering. Then it was put in the scene and brought into the Flame, where we pre-built the background. So we gave it that feel of a book and that texture of a book. And then we would composite in the models normally, and do everything you would do to make a composite look right, so that they were differentiated but they didn’t look like a bad composite. They maintained this soft, realistic look after we hypersharpened the paper itself to give it that edge, that crispiness.
We used depth passes and depth maps to create a shallow depth of field – we were playing a scale game with it. You’re shooting a miniature book, but the person’s apparently real-life. Is it a human-sized book or is it a small person? We didn’t want to give you an answer to that. We wanted to make sure it felt like a macro book with a real live person in it. We added a little more depth than you would have if it was a human-sized book.
What was the most unusual aspect of the job?
The real challenge for us was just the data-wrangling. The job had so many threads going in so many different directions. We had four weeks from shoot to deliver. We had four :15s, two :30s, and two other :15s that were text versions. I’d have a CG artist that couldn’t model yet because he didn’t have pictures, so I’d send him out to take pictures. And then he comes back to do his pictures while I’m cutting things out. We had to have that cohesive teamwork. We really needed that to be able to deliver on time – and give me and the director and the clients enough time to sit down and make sure everyone likes the look and is happy with the project. We were dealing with a director who’s an FX director, so that adds another layer, too. He speaks the language and he knows what he’s talking about. It helped us in so many ways – he knew, “Hey, let’s do a pre-vis on this and stick to it.” That eliminated that variable for us, so we could concentrate on the look.
Did you enjoy this article? Sign up to receive the StudioDaily Fix eletter containing the latest stories, including news, videos, interviews, reviews and more.
