Tilting At Windmills: The Sometimes Quixotic Pursuit of Transferring Digital Video to Film
At this year's DOCtober, 19 of the 21 films screened were shot on Digital Video (DV) and blown up to film. At the recent AFI Fest, 11 of the 12 docs in competition were shot digitally and transferred to film. The digital future is here—still imperfect, but readily available and getting more affordable. In my ongoing effort to conquer a case of technophobia, I decided to investigate why some prints that originate from DV look so much better than others. An illuminating case study is that of Keith Fulton and Louis Pepe's Lost in La Mancha, the unsettling and compelling story of the "unmaking" of Terry Gilliam's The Man Who Killed Don Quixote.
Like many documentarians, Fulton and Pepe weren't thinking about blowing up to film when they started shooting. Because their financing was initially for a broadcast hour, they shot on an NTSC Sony PD-150, with a standard 1:33 aspect ratio—"Everything we wouldn't do if we'd known we were shooting to blow up to film," says Fulton. "Ideally, we'd have shot on PAL, and framed for 1:85, using an external anamorphic adapter." "Well, ideally, we would've used the new Panasonic 24p camera," adds Pepe.
Fulton and Pepe had learned from their last doc how to maximize shooting in video, which ultimately served them well in the blow-up process. Pepe shot La Mancha mostly in medium shots and close-ups. "We tried not to lose color information," he says. "In DV, there's color compression, so when you're zoomed out to the maximum, the edges get muddier." When Pepe did shoot wide shots, he tended to place a large object or person in the foreground, so the frame felt more forgiving of any jags in the background. He also avoided pans as much as possible to eliminate strobing, and always shot with a wide angle adapter. "It reminds you to get close to your subject, and it just looks more dynamic," he maintains. "Of course, it minimizes camera shake, which gets enlarged in a blow-up. And focus is paramount for a blow-up, even over image quality." While shooting Gilliam and his crew watching dailies in a darkened room, Pepe chose to dial up the gain, opting not to open up the aperture. "I figured it was better to have an image that was grainy or noisy than to be out of focus," he says.
Lost in La Mancha was cut on Final Cut Pro 2.0. Once editor Jacob Bricca got close to locking picture, Fulton and Pepe did tests with several post houses and labs. They settled on The Post Group in Los Angeles for the lengthy online, and Colour Film Services (CFS) in London for the film print. The pair did exhaustive research with both facilities to map out the process ahead of them. Fulton explains, "It was an online that required a serious flow chart. We'd done lots of onlines before, but some of this work was new to us, and we were learning as we went." CFS gave them a full list of issues to address before they went into the online and the Post Group had a detailed spec meeting with the filmmakers to identify potential issues and problems. Pete Nehl, tech supervisor at the Post Group, (title??) says this initial meeting is critical, "especially for financially-challenged documentarians." "At this point, we had to pick between a 1:85 or 1:66 aspect ratio; it was a touch choice," explains Fulton. They chose 1:66, to get the most height possible into the frame.
The team began with multiple elements for the online: an output from FCP to DV-Cam of the locked picture; output from FCP of selects with handles (shots which needed special treatment like speed changes, digital zooms, etc); and animation sequences on TIFF files—all of which were bumped up to NTSC Digi-Beta. A fourth element, original material shot on Gilliam's PAL camera, was bumped up to PAL DigiBeta. All of the PAL elements were converted to NTSC, and then assembled in an NTSC DigiBeta online textless master.
CFS advised the filmmakers to do as much color correction as possible on the video, to avoid the cost of doing so on the film print. Fulton and Pepe made a list of scenes they knew needed tweaking, and sat with a colorist on the Da Vinci for only four hours, due to their limited budget. At this point, they began a two-track process, to create both an NTSC and a PAL master; the PAL would be used for the film blow-up. "The quality of the color correction and of the PAL version—the standards conversion—is a key part of how good your print will look," Fulton explains. "We'd done a test at another lab, which also used the Alchemist, but it didn't look very good. It was soft and washed-out looking. The Post Group's work was really superior." Pepe adds, "It turns out that a standards conversion is not a fixed entity; it requires a human factor, a human eyeball to monitor the transfer." The Post Group's Nehl explains that the Alchemist's technology isn't proprietary, but that there are variables that are programmed by a facility's specialists—filtering, enhancement, etc. Different algorhythms can be used to "de-interlace" the video, and some scenes work better with certain algorythms than others. "It actually takes an aesthetic sensibility to do an excellent standards conversion," says Nehl.
At this point, Fulton and Pepe went back to the original PAL DV-Cam footage, bumped it up to PAL Digi-Beta, and cut it back into the new PAL version. They did another color correction, this time on the PAL version, then did an extensive titling session on both the NTSC and PAL versions. They were now almost ready for the "film out master," but there were two more considerations: they left on the lower thirds, but took off all subtitles; and they had to deal with framing for 1:66. Fulton and Pepe watched the entire film on a monitor masked for 1:66, and identified about 20 areas that they wanted to reframe for the 1:66 print. Lastly, they had to decide between mastering onto Hi-Def or Digi-Beta. Though some labs prefer Hi-Def, CFS advised mastering onto Digi-Beta to soften the sharpness of the video.
Because the film print would be made from the PAL master, its length would be four per cent longer than the video master—the result of going from 25 to 24fps. One last consideration was for the end titles. CFS advised on the font, font size, and speed of the crawl. Most labs doing tape-to-film work suggest that, for the best quality end titles, avoid a crawl and do cards, or originate the end titles crawl on film.
After two weeks, Fulton and Pepe had an NTSC master for broadcast in the US and a PAL film out master, partially textless and reframed for 1:66, with original PAL footage cut in and color corrected, and with the audio partially pitch corrected. At this point, the master was shipped to CFS in London, and producer Lucy Darwin and the lab took over. Due to budget limitations, the first time Fulton and Pepe saw a print was at the premiere at the Berlin Film Festival, before 750 people! Once they returned to the US, they took notes on the print, reel by reel. Before more prints were made, the filmmakers were able to make some changes to brightness and tint. Again, due to the budget, changes could be made only on an entire reel.
Even after release prints were made, the filmmakers' work was not done. "You have to be thinking of exhibition factors, where you don't have any control, and figure out how to anticipate and minimize problems and errors," cautions Pepe. "There are still a lot of places that don't project 1:66, so it's all about alerting people to that. Usually festivals are equipped with various lenses for the different aspect ratios, and sometimes art houses can even show it at 1:33, since they show old movies at that ratio. But it's not what many projectionists are used to, as features are released in 1:85."
"Whenever we can, we show it at 1:33," Fulton explains. "But there's still projectionist error, and sometimes the film cans are incorrectly labeled 1:85 so the film is projected too tight. We've shown up at screenings where the film was shown at 1:85, and we'd cringe when people would tell us that some subtitles were barely readable. If the projectionist frames to the academy leader, it's fine, but that's not always done. The negative is full-frame, so there's a lot of room for error."
"This whole technical morass was because we were shooting on DV," Fulton continues. " So why shoot on DV? Because DV is the natural evolution of what the direct cinema pioneers were doing. They were all about streamlining the process—going for camera portability, shooting handheld and working with a one- or two-person crew. The whole goal of observational cinema is to be inobtrusive," Pepe adds, "And DV can foster intimacy; the technology has a huge impact on the relationship between the filmmakers and subjects, which impacts the storytelling. And ultimately, it's all about storytelling."
Their advice to other filmmakers? "Most doc-makers don't start off knowing they'll be blowing up; that decision usually comes when you realize you've got the footage to make a feature-length doc, one with theatrical potential. So, do the best you can with what you have. Most things can be fixed later. The focus has to be on getting a good story first!"
On my last Web search, there were over 15 labs specializing in DV-to-film transfers. Many of the labs have websites with informative FAQs and guidelines for production and post-production. If filmmakers think that blow-up to film from DV acquisition is an option, before they begin shooting they should consider different cameras and settings—size of chips, PAL vs NTSC, lens, shutter speed—aspect ratio and shooting style. Critical post-production decisions need to be made during the online and color correction—don't crush levels, identify footage with special needs (archival, moves on stills, effected shots, animation, titles). A high-quality standards conversion is key if you're going to PAL; do tests on standards conversion and a blow-up. Many labs will credit back some of the cost of a blow-up test if they do the final film print.
And with all of this said, the future of DV to film may change drastically with the introduction of the new 24P cameras. While Hi-Def 24P has been around for awhile, just a few months ago Panasonic released the DVX100, a Mini-DV 24P camera. With one software addition to Final Cut Pro, filmmakers can now shoot video at 24 fps in progressive scan, and end up exporting from FCP an NTSC color-corrected video that is ready to blow-up to film, with no standards conversion or length change to worry about.
Lisa Leeman is currently directing a documentary being shot on PAL Digi-Beta and DV-Cam. Researching this article did help dissipate her technophobia.