July 2000

Amargosa, a 90-minute, indie documentary directed by Todd Robinson, details the passions of dancer/performance artist, Marta Becket, who is now in her 70s. One day, for no apparent reason, Becket decided to transform an abandoned theater in Death Valley, California, into her personal performance venue, the Amargosa Opera House. She continues to perform there to this day.

As fascinating as Becket's saga is, the story of how filmmakers transformed the project from its original 16mm state into a 35mm blowup is almost as compelling. In early 1999, the filmmakers, on a limited budget, decided to shoot on Super 16mm Kodak Vision stock. They shot for several weeks, not knowing at that point whether the film's primary distribution medium would be film or television. They later decided that they would submit the film for Academy consideration and hurriedly began editing and posting so that they could make the October deadline. They did, and the film later premiered at Slamdance. But during their flurry of activity, filmmakers did not consider how or where they would perform the 35mm blowup. They ended up transferring the film to DigiBeta as a video master in NTSC without considering that this might not be the best format for the project's eventual film distribution.

"We decided to produce a video master, which we thought gave us a lot of advantages in terms of editing, color correction, and optical effects, like the many dissolves in the film," says producer Sidney Sherman. "But we hadn't fully researched how to make a 35mm print out of that video master. In retrospect, we should have researched that issue thoroughly before transferring even a foot of film to video."

After several tests, the Amargosa team eventually brought the film to Hollywood-based Cinesite. Filmmakers chose Cinesite because the company could deliver a 2K resolution blow-up. The only difficulty was that Cinesite's method of obtaining this blowup required a more film-friendly (25fps as opposed to NTSC's 30fps) PAL video master. "That's when we realized we would have been better off transferring the original film to PAL," admits Sherman. "Now, we would have to address the 3:2 pulldown issue."

Cinesite does not routinely perform video-to-film conversion and normally would not have even attempted the project. However, at the time, the facility was coming out of Pleasantville and was preparing to undertake experimental electronic color-timing work on the new Coen brothers' film, tentatively titled Oh Brother. The company was also pondering moving into the film-conversion business (a move that it later put on hold). So Cinesite took on Amargosa to test its conversion pipeline's capabilities. "We wanted to see how much memory it would take to scan in an entire feature on video and then record it out to film," explains Bob Fernley, manager of scanning and recording. "We wanted to test our ability to do such work before moving on to the Coen brothers' film. We got a lot of R&D done with this project."

First, to convert Armagosa from NTSC to PAL, Sherman brought the video master back to Burbank-based International Video Conversions, which had performed the original dirt-fixing work. Using its TK 3:2 conversion process (a technique specific to film-originated material), IVC created a PAL D1 video master with reel breaks for Cinesite.

Cinesite then captured the video in its SGI Octane computers and brought the digital information into a Kodak Cineon workstation (version 4.1) to separate each of the video frame fields into 24fps format and crop them for the Academy's standard film-projection aspect ratio (1:8:5). The big problem, however, was still the 3:2 pulldown issue. Certain frames suffered from severe ghosting problems that Cinesite had to tweak in the Cineon environment. "For those frames, we re-captured the files, brought them back into Cineon, separated them, and re-combined them in a slightly different way," says Fernley. "Rather than doing a 50-50 dissolve, it was more of a hit-or-miss game as our operator just kept tweaking and re-combining the fields to improve the images and reduce the video artifacts. There is no way to eliminate 100 percent of the artifacts, but we were able to significantly reduce them to the point that only a trained eye would ever notice them on the big screen."

After handling those issues, Cinesite recorded the film frames out in sequence to its Cineon Lightning recorders and provided filmmakers, at last, with a 35mm negative. The Armagosa team took that negative to CFI, Hollywood, to make prints. Although P.J. Marsiglia of Santa Monica-based Company 3 had already color corrected the video master during its original transfer to NTSC, CFI's Chris Regan performed additional color-timing work while creating the answer print. Sherman notes that this additional work was necessary because video going out to film often loses some of its early color-correction and color-timing subtleties.

Cinesite dedicated numerous painstaking hours to the small, indie project, at what Fernley calls "a favorable price point." But, he says, the experience was worthwhile because it permitted the company to experiment with the huge amounts of data involved in scanning and recording features. And for the filmmakers, the experience taught them both about the film transform process and the importance of making crucial post decisions during preproduction and production.

Amagosa's DP, Curt Apduhan, reflects on the project: "It was a great education for all of us. Unfortunately, much of that education came after the water had already left the pail. Had we known when we started shooting what we know now, we would have probably transferred the film to PAL and cut negative, which would have allowed us to easily conform the piece for any format later on. Even earlier, if we could have found the money, we would have shot 35mm to begin with. But obviously, we had limited money, so we shot 16mm and tried NTSC. I'd advise indie filmmakers to always presume they will start distribution on the film side, and if they can afford it, shoot 35mm to begin with and cut negative. That's hindsight, though, and not always possible. With high-def and other things coming along, the problems we encountered may not be big problems in the future."