The American Society of Cinematographers

Loyalty • Progress • Artistry
Return to Table of Contents
Return to Table of Contents April 2013 Return to Table of Contents
Oz the Great and Powerful
Page 2
Page 3
To the Wonder
Presidents Desk
ASC Close-Up

Peter Deming, ASC reteams with Sam Raimi on Oz the Great and Powerful, the prequel to a beloved Hollywood classic.

Photos by Merie Weismiller Wallace, SMPSP. Photos and frame grabs courtesy of Walt Disney Pictures.

Adapted from L. Frank Baum’s series of books about the magical land of Oz, the new film Oz the Great and Powerful follows young circus magician Oscar Diggs (James Franco), whose life is transformed after a sudden, violent storm sends his hot-air balloon hurtling into the parallel world of Oz. There, he encounters three witches, Theodora (Mila Kunis), Evanora (Rachel Weisz) and Glinda (Michelle Williams), who believe his destiny is to become the wizard who will save Oz from the tyranny of the Wicked Witch.

The film reunited director Sam Raimi and cinematographer Peter Deming, ASC, who last collaborated on the hyperkinetic horror film Drag Me to Hell (AC June ’09). Both filmmakers were new to 3-D digital capture, and in prep, they tested both the Red Epic and the Arri Alexa with stereo rigs. They opted to shoot with the Epic at 5K and 4:1 compression (for a final 2.40:1 extraction) largely because of the picture’s extensive visual-effects work. Deming explains, “The plan was to release a 2K DCP, and this [capture method] enabled the visual-effects team to punch in on frames by up to 50 percent without any resolution loss at all.”

Deming’s research included attending a cinematography workshop at Sony Pictures’ 3-D Technology Center, and he also surrounded himself on set with some seasoned 3-D vets, including stereoscopic supervisor Ed Marsh and main-unit stereographer James Goldman. “Between the two of them, we covered all the bases involving theory, practice and hardware,” Deming remarks.

Two Element Technica Atom 3-D rigs were assigned to the main unit, with an additional one on a splinter unit led by cinematographer Paul Sanchez. Each rig employed a 3mm-thick beam-splitter mirror, which required a 1-stop exposure compensation, as well as a custom-size Quarter Wave Retarder, which A-camera 1st AC David Eubank describes as “a filter that corrects polarization mismatches between the left-eye camera and the right-eye one. This imbalance could be glare from backlit images or windows, or subtle differences such as highlights on an actor’s face.”

Deming shot the entire picture with Primo prime lenses supplied by Panavision Woodland Hills. “We shot primarily on the wide end of our lens choices,” he says. “We rarely went to a 50mm except for extreme close-ups, and we only used a 75mm or 100mm for inserts. Most of the movie was shot at 40mm or wider. This was a big adjustment for me, particularly when shooting coverage of beautiful women. We were sometimes at 27mm for a close-up, and you’d never do that on a 2-D picture!

“There was something about the 3-D image that was really forgiving,” he continues. “Sometimes we would pull back if we wanted to minimize the distortion, to keep the image flat in terms of facial features, knowing that we could punch in during post without losing resolution. At a focal length of 27mm, the interocular distance contributes to expanding or reducing the roundness of faces, and you can really manipulate the size and shape of round objects in 3-D.”

The filmmakers shot parallel stereo on set and converged in post mainly to help simplify Sony Pictures Imageworks’ visual-effects work, which was supervised by Scott Stokdyk. CG shots are often complicated by the image distortions, keystoning and disparities caused by toeing in the cameras for converged capture. When shooting parallel stereo, an interocular distance is set between the left and right eyes on set, and the convergence effect is achieved in post by offsetting the left and right eyes on the horizontal axis. Deming was careful to add a 10-percent pad to the edges of the frame “because when you converge in post, you lose part of the shot. You can’t assume you will have the whole image.”

“Our 5K image was usually down-sampled to 2276x1138,” says Marsh. “That gave us 114 pixels on each side of the image that could be used for convergence, as well as additional information at the top and bottom. When we didn’t need it for convergence, we still had the ability to reposition images quite a lot, which often helped during the editorial process.

“When you shoot parallel, you still need the ability to converge the image for viewing on set; otherwise, everything appears to be forward of the viewing plane, and that simply isn’t comfortable or artistically correct,” Marsh continues. “We used Raptor to converge our image feeds on set without baking that choice into our raw camera files.”

Goldman notes that a few shots were converged during capture. “They were shots that would have been a little tougher to achieve in post, like when the Winkie guards thrust their axes at the camera,” he explains. “These were the bigger 3-D moments, and Sam and Peter wanted to make sure all of the elements were right where they were supposed to be in the shot.”

Notes on lens sizes, convergence and interaxial settings were downloaded to a Filemaker Pro database designed by Marsh. Metadata was transmitted from the rigs through the Atom’s 3-D IO module to a MicroSD card, which was then downloaded and integrated with the digital dailies created specifically for the Avid. Dailies were viewed in a screening room. “Our dailies had our convergence choice baked in, but on shots where that convergence point needed to evolve with the shot, it was a bit problematic,” says Goldman. “No one has a system to track convergence and re-create it after the fact when you’re shooting parallel. You can slide stuff around with key frames in the Avid, but it’s a lot of work to get it where you want it.”

When loaded with two Epics, two lenses, the mirror box and a wireless Cmotion Cvolution focus-control unit, each rig weighed almost 85 pounds. “We weren’t going to subject anyone to handholding that thing,” quips key grip Phil Sloan. The size and scope of the sets mandated a 50' Technocrane as the show’s main camera support. A-camera operator Patrick Rousseau was assisted on the Technocrane by dolly grip Tim Collins and crane operator Derlin Brynford-Jones.

Oz was shot in Pontiac, Mich., entirely onstage at Michigan Motion Picture Studios (known at the time as Raleigh Studios), where the production occupied all seven stages. “Our stages were literally wall-to-wall with set,” says Deming. “Size wise, our sets were probably on par with those of The Wizard of Oz!” During a round of additional shooting in Culver City, a Yellow Brick Road stage was set up at Sony Studios, located on the former site of MGM, on the same stage where cinematographers Hal Rosson, ASC and Allen B. Davey, ASC filmed the original Yellow Brick Road. “That was a pretty cool coincidence,” says Deming.

At Raleigh, four of the production’s seven stages were completely encircled in bluescreen from floor to perms (45'), and a fifth was three-quarters encircled. This facilitated the construction of large practical sets that the Imageworks team could then extend in any direction with CGI. “We chose bluescreen instead of greenscreen because there’s a lot of greenery in this movie, and there’s also an Emerald City,” says Deming. “Plus, I think we would have gone insane looking at all that green for six months!”


next >>