TMCnet News
Scalawags in Stereo [American Cinematographer, The](American Cinematographer, The Via Acquire Media NewsEdge) Dariusz Wolski, ASC tackles 3-D capture for Pirares of the Caribbean. On Stranger Tides. Looking buck, Dariusz Wolski, ASC chuckles ut his Steadiaim operator's plight while trying to capture Johnny Depp's performance on an isolated beach in Fajardo, Puerto Rico, for - climactic scene in Pirates of the Carrebean On Stranger Tides. Because the movie was being caprured in native 3-D, Stcadicam operator David Luckenbach was laboring beneath a heavy stereo rig, waiting tor director Rob Marshall to call "action." "The ng was so heavy, and we were on sand," Wolski recalls. "David could feel his feet sinking all the way to his ankles. He couldn't lilt his foot when they called 'Action,' so he constantly had to stomp up and down to keep his feet tree. That's an example of how exhausting this shoot was." Marshall calls the efforts of WoLskis team "Herculean," noting that Pi m fes is "the most physical film" he lias ever directed. "The story required things that are difficult to accomplis!! with 3-D rigs," he observes. "Fortumitely, everyone was up for the adventure." Disney's decision to shoot the latest installment in its Pirates franchise in 3-D makes die film a high -pro file test ease for the latest paradigm shut in stereoscopic cinematography: bringing it out of controlled environments and into the real world - and, in this case, to a number ot rugged locations. "We felt like pioneers," says Marshall. "We were working on beaches, in caves and on boats in places like Hawaii, Puerto Rico and London, and doing it with sensitive cameras and equipment," continues die director. "Some of the areas were so remote, we had to bring in gear by helicopter. I give a lot ot credit to Darius and all the experts who helped us." Wolski was a veteran of the Pinito franchise, having shot all three ot the previous pictures, Toe Curse of the Black Pearl (AC Aug. '03), Dead Man's Chest and At World's End (AC June '07), and he had also studied digital cameras and 3-D rigs while prcpping Alice in Wonderland for Tim Burton (AC April '10). In the end, Alice was shot in 2-D and converted to 3-D in post, but Wolski 's research into the Pace/Cameron Fusion 3-D ng during that period made him comfortable using it on Pimks. Additional testing made it clear that ol all the digitalcapture systems available at the time, the Red One with the Mystcrium-X sensor was his best option in terms ot size, weight, compatibility with cinestyle lenses, ability to record to solidstate media (16-gigabyte Red CF cards), and 4K. resolution. "At the time, the Arri Alcxa wasn't available, and die Sony cameras we tested were too heavy," explains Wolski. "The Red cameras with the MX chip were more manageable. Tl icy were die best option considering that we were committed to going to faraway places and shooting the way the [previous] films had been shot. The challenge became how to adopt the 3-D technology and make it flexible enough for a movie of.' this sede." Wolski has a longstanding relationship with Panavision, so he wanted that company to service the production, even though he was using Red cameras and Pace rigs. He therefore asked his first assistant, Trevor Looinis, to bring Panavision and Red together on the project. Panavision ended up purchasing anil prodding 13 Red cameras to the main unit and six to the second unit, plus multiple sets of matching lenses. Red also provided support in partnership with camera-data supervisor Jerocn Hendriks, particularly in terms of writing cu jm firmware for the Red camera that tailored them to 3-D capturt "We essentially brought Panavision, Red and Pace together, and they all helped make it happen," says Loonns. Panavision's lens expertise proved to be critical. Senior technical adviser Dan Sasaki sorted through more than 160 lens pairs to find the best matches, ultimately culling 76 pairs tor the show. Paiiinision provided three sets of Zeiss Ultra Primes, two 15-40mm Angenieux Optimo zooms, two 28-76mm Optimo zooms, a set of Primo Close-Focus primes, a set of Primo (14.5-50nim) Macro Zooms, one Primo 4:1 (17.575mm) zoom and one Primo 11:1 (24275rnm) zoom. The first unit shot mostly with the Ultra Primes and Optimos, reserving the Primos for 2-D capture, wliich amounted to underwater work, extreme close-ups and some high-speed shots, according to Wolski. The second unit, led by director of photography Patrick Loungway, used Ultra Primes and Optimos on Fusion rigs alongside the first unit in Hawaii and Los Angeles, and used Primo primes and Optimos on Element Technica stereo rigs in London, says Loungway. "We liad to establish limitations and tolerances tor allowable focallength disparity, concentricity of the lens mount to the sensor, and overall] compatibility of lens-focus scale to matched lens sets," Sasaki explains. "We tried several methods ot finding matches initially, such as wedding a microscope to the projection bench, using the MTF [Modulation Transfer Function] to give us point-spread readings, and mapping the lenses on the projection bench at like magnifications. In die end, using the projection bench with respect to lens MTF readings gave us the most consistent results. We found that we had to find focal-length matches within .25 percent of each other, as well as maintain the overall quality ot the lens image." Wolski says the short zooms were especially useful on the 3-D rigs because ot their light weight and small size. They also permitted him to shoot wide open, the style he prefers for 3-D capture. "Once we had the zoom on die camera, we couldn't zoom during the shot because Lit] wouldn't track correctly with die two lenses, and there was no way to have them totally in sync," says Wolski. "However, we learned that once we zoomed in, we could quickly realign [the figs J, so we ended up using die zooms more as variable primes. That saved us some time in lens changes; it rook about iour minutes to change them out, which is redly fast tor 3-D." Another key issue involved which method to employ to set the interocular (a.k.a. inter-axial) distance and convergence point between left-eye and righteye imagen'. Wolski notes that many native 3-D projects rely on mechanically setting and controlling the convergence point on set, and it's often set to correspond to a point in focus in the main action, or what is called the "screen plane" or "stereo window." (The convergence setting doesn't always have to relate to die same point as where the lenses are focused, however.) But because Red's MX sensor can record 4.5K-rcsolution imagery (in its 4.SK widcscreen 2.35:1 mode), Pini/es visualeffects supervisor Charlie Gibson suggested that the production should not set the convergence point on set, but instead simply shoot parallel and leave the convergence work tor post. By doing so, the tilmmakers conici preserve extra pixels around the extraction aira by avoiding distortion in closer objects; such distortion can happen when images are mechanically converged incamera. (3-D expert Rob Engle of Sony Pictures Image works also ottered advice in prep.) "Converging in-camera was common before |this production] because ot the limited resolution ot the digital cameras that were available," explains Dave Drzewiecki, who sen'ed as the stereographcr for most of Pirales. (James Goldman, Wolski's longtime second assistant, took over near the end ot the show when Drzewiecki moved on to another project.) "Back then, you didn't have those extra pixels around the images available to use tor realignment later, anyway," continues Dr/ewiecki. "Whenever you do an image adjustment to correct an error or to create convergence [on setj, you sacrifice pixels and, theretore, resolution . But with the greater resolution of the Red MX sensor, we could take a 4K slice out out the middle oí the 4.5K nnage in post and designate pixels around it left-to-right as realignment pixels. So shooting parallel made sense." It was Hendriks' iob to quali tycontrol the recorded images on set hy viewing, syncing and backing them up. This included examining stereo quality by building rough VD tiles on location - applying a 1 -percent offset to synced imager\r using Assimilate Scratch powered by Red Rocket accelerator cards. As the production moved around the globe, the iilmmakers sent .VD imagery and metadata from the set ro Company 3s headquarters in Santa Monica and to Deluxe 142 in London, using a remote data center set up at each location hy Company 3. Various coloristi graded the dailies and then sent them on to editorial, also sending those same color settings as metadata back to the location, where Drzewiecki would apply them to tiles on site. (Wolski also did the final color correction at Company 3, working with ASC associate member Stefan Sonnenfeld.) But the process started with data downloads and checks on location. Hendriks used Scratch on a mobile 3-D station in tandem with 1 Bevond's Wrangler Pro and RedCine-X software tools, which enabled him to do Rillresolution real-time checks ot the stereo images. His 3-D station was little more than a small suitcase with a hard drive, a flip-up keyboard and a 24" LCD monitor. "I would download information from the camera to my internal drive and build 3-D files out of it," Hendriks explains. "My job was to make sure that we knew exactly what we liad before we left a location. During even' reioad, about every nine minutes, Fd use this system to check everything - to see it time code matched, to check for any gen-lock problems, and to make sure we had both eyes to build a 3-D file. I did that for every shot as we went along, and I backed everything up even before we sent the cards to our data center." The stcreographers and Fusionrig technicians, meanwhile, were often busy realigning rigs. Even with constant vigilance, alignment positions between left and right cameras could move slightly during routine handling, temperature shifts and the like. To align and track rigs and lenses, and to balance color, the stereographers used two 3ality Digital Stereoscopic Image Processors. Drzcwiccki calls the SIPs crucial and "essentially a default standard for how to capture 3-D in terms of analytical information about the 3-D image. They allowed us to have a level of alignment precision that we could nor do visually using traditional methods." Hendrikus and Drzewiecki also devised a solution for viewing 3-D on location on something other than small monitors. They paired 3-D beam-splitting techniques with 22" LCD panels from Cinetronics, configuring two panels and electronic components together into a cube-like monitor that allowed the crew to view full-resolution Linages with passive 3-D glasses. Despite such innovations, shooting a large-scale action -adven ture film in stereo was still a grueling attain Capturing stunts and other second-unit work was particularly difficult, largely because the nature of 3-D acquisition prevented die team from using many of the rime-honored tricks for such work Loungway notes that the ability to be spontaneous while shooting action scenes, a freedom the tilmmakcrs had enjoyed on the previous Pirates movies, was curtailed by the 3-D rigs. In effect, the camera footprint had to he more invasive this time around. "For the first three films, the filmmaking presence was sometimes just a handheld camera with a couple ot guys on a large boat or set with a small crew," says Loungway who was the secondunit cinematographer on Dead hLuù Cbesf and At World's End and did additional photography on The Curse of the Black Pear!. "With 3-D capture, the ability to shoot jaxx- style, oft- the -eu tir, wasn't possible. VVc did have spontaneity, but within the context of having to figure out how to do it in the third dimension. Spontaneous decision^ sometimes proved to be impractical." In particular, whip-panning fast enough to record action like the carriage chase through London proved to he complicated because, as Loungway notes, "a 100-pound rig, once moving, doesn't want to stop quickly. But we had solutions thiit worked well; we used Libra and Chapman G3 remote heads a lot on second unit. We might not have needed then: it this weren't a 3-D movie." For a few specific shots, he continues, the second-unit team even hung 3-D rigs from an 80' Condor on bungee lines to swing the rigs around in a 1()'-15' area. It was an old-school technique he and Wolski were inspired to try after Lounuway recalled an American Cititnnaiogmpbei' cover photo of legendär}' cinematographer Jack Cardiff, BSC swinging a 3-strip Technicolor camera on ropes tor Th? Red Shoes (AC Dec. '48). "If you think about it, they were handholding a giant camera, in a sense," Wolski says of the Cardiff photo. "In other words, the size of the camera didn't stop them. That's the attitude we had on this project." Indeed, the filmmakers made extensive use of Condors, Technocrânes, and various other traditional tools to surmount obstacles. A big visual-eftects sequence, for instance, features an attack by an army of mermaids at night. Establishing shots were captured at I lalona Cove in Oahu, but the waves were so treacherous that the filnumkers liad to replicate the cove on the lot at Universal Studios to shoot required plates. Loungway says most of this work on the Universal lot was shot using Technocranes on barges, not only for sweeping crane shots, hut -also tor dramatic pans. In I IaW1Ui, the production also had to shoot a major night-exterior sequence on a galleon at a location called Kaneohe Buy. The filmmakers encountered high winds on the night of the shoot, threatening gaffer Rafael Sanchez's plan to provide subtle moonlight. "We had balloons on cranes and were trying to keep the company moving, but we had to minimize the dumgc to the equipment i a those high winds," recalls Sánchez. "In the end, we were able to wrap the balloons in clear Grirrolyn.That didn't cut any light irom the balloon, but it protected them and added another layer to help disperse the wind around them." Overall, Sanchez rmaintained the same ligli ting techniques he used on the previous Pirates films. "Our approach was mostly traditional," he says. "In keeping with all the Pimtes mcnics, we were idwuys trying to motivate as many practica! sources as possible - lots ot warm light for firelit and candlelit interiors, and cool moonlight tor night work." Thanks to thorough preproduction testing, he and Wolski were confident that this approach would hold up. "Our conclusion was that u single Red camera responded incredibly well in extremely low light, but the two cameras and the mirror [in the Fusion rig] would reduce things by about one stop," says Sanchez. "We went from about 800 ASA, which is the desired ASA for Red, to 400 ASA, which is closer to 35mm. That meant we could light things pretty much as we always had." Bearn-splitting mirror technology, of course, lies at the heart of the optical system in modern 3-D rigs, and for Pi rafes, the production chose the most sophisticated mirrors Pace offers - dubbed "organic mirrors." Wolski and Drzewiecki say they believe these are currently the least-polarizing mirrors available for 3-D rigs. That's important because polarization can introduce specular anomalies and color shifts into the recorded imagery from the reflected eye, creating subtle differences betumeen lefteye and right-eye images that cannot always be corrected in post. The organic mirrors are particularly delicate, and the production went through more than 30 of them. "The Pace mirrors produce beautiful images, but they arc incredibly fragile, and that makes them susceptible to moisture and salt aii'," says Drzcwiccki. "As soon as we took them out of their wrapping in these environments, they would start to degrade. Sometimes we went through one or more in half a day, depending on conditions. We were putting them in the worst possible environments, su it was costly. But you could argue it was money well spent because it all shows up on the screen." For the crew and cast, there were also hurdles to overcome in terms of learning how to use framing and blocking to achieve desired compositions in 3-D. All these issues, says Wolski, are why the production was "experimental" and "risky." (Indeed, the experimentation went on until production's final moments, as the filmmakers circled back ior pickups using Red Epic 5K prototypes with Element Technica and Paradise 3-D rigs in addition to Fusion rigs.) But, VVoJski adds, Pirates to benefited significantly from timing, in the sense that stereo -capture techniques and workflows were rapidly advancing throughout the shoot. "To shoot a movie like this in 3-D when the workflow is not mature and the technology is changing, you have to combine everyone's expertise to achieve your goal," says Wolski. "You need people who can put all this knowledge together efficiently, and we had an expert team doing that, which enabled Rob Marshall and me to focus on the story. The most important lesson is that we need to dcmythologize 3-D. It's still filmaking." Opposite: Capt. Jack Sparrow (Johnny Depp, center) joins a harrowing search for the Fountain of Youth in Pirates of the Caribbean: On Stranger Tides. This page, top: Blackbeard !an McShane) encourages Sparrow to cooperate. Middle: Capt. Barbossa (Geoffrey Rush, second from left) takes up pursuit. Bottom: Cinematographer Dariusz Wolski, ASC checks the 3-D rig fitted with Red One cameras on location on Palomino Island, off the coast of Puerto Rico. Right: The crew captures a shot inside the Painted Hall at the Old Royal Naval College in Greenwich, Engfand. The location stood in for St. James Palace, where Sparrow is forced to meet with King George. Below: Sparrow exits St. James Palace in his inimitable fashion. Clockwise from left: In his flight from King George and his men, Sparrow commandeers a coal wagon: the second unit follows the chase from a dolly; another 3-D rig follows the action from above. Clockwise from right: Blackboard's sinister ship, the Queen Anne's Revenge, sets sail for the Fountain of Youth; Blackbeard practices the dark arts in his cabin; a crane follows the action aboard the ship. Top: Working on a set constructed on the Universal Studios backlot, the crew prepares to roll camera on an attack by an army of mermaids. Middle: A camera housing is employed to get a closer angle on the mermaids. Bottom: The crew sets up a night shot on location at Oahu's Halona Cove, where some of the action around the "mermaid pools" was also filmed. Right: Technicians at work at the 3-D monitoring stations set up during filming at Pinewood Studios in England. Below: For location work in more rugged terrain, camera-data supervisor Jeroen Hendriks worked with a mobile 3-D station to view, sync and back up the recorded stereo images. Director Rob Marshall talks with Depp on the Fountain of Youth set, which was buift on the 007 Stage at Pinewood Studios. Marshall says Pirares is "the most physical film" he has ever directed, and praises the "Herculean" efforts of his collaborators. "With 3-D capture, the ability to shoot jazz style, off-the-cuff, wasn't possible. Spontaneous decisions sometimes proved to be impractical." Wolski and second-unit cinematographer Patrick Loungway (far left) line up a shot on location outside the Old Royal Naval College. TECHNICAL SPECS 2.40:1 3-D and 2-D Digital Capture Red One, Epic Zeiss Ultra Prime, Angenieux Optimo, Panavision Primo Michael Goldman is a freelance writer and the author of the blog "The Art of the Craft" at www.theasc.com ("Scalawags in Stereo," p. 26). (c) 2011 American Society of Cinematographers |