Dust to Glory is director Dana Brown’s latest action-adventure documentary. It sucks the viewer right into the middle of the Baja 1000, the longest point-to-point race in the world. Attracting hundreds of participants of all abilities, from amateur dune-buggy drivers to motorcycle supercross legend “Mouse” McCoy and racing icons such as Robbie Gordon and Grand Marshall Mario Andretti, the race, the competition and the camaraderie are all chronicled in this magnificent documentary.To capture the 40-hour event—a grueling, unpredictable off-road race through the desert of Mexico’s Baja Peninsula—the crew of over 80 used 55 cameras, four helicopters and a four-passenger buggy camera car. The formidable scope and complex logistics of production required a dogged determination by the filmmakers that matched the sheer determination of the race participants.But the challenge did not end at the finish line. Postproduction was equally adventuresome. With over 250 hours of dailies in just about every imaginable format—35mm, super 16, HD, DVCam, NTSC, night vision and mini-DV—the project morphed into a postproduction marathon.Brown and producer/editor Scott Waugh combed through the massive amount of footage for the moments and stories that would become the final 98-minute movie. It fell upon online editor and supervisor Jacob Rosenberg to technically assemble the pieces into a visually cohesive film. He chose an all-digital HD workflow—the newly introduced PC-based, desktop system, Adobe Premiere Pro 1.5 and its companion program, the tried-and-true Adobe After Effects 6.5—to mix media, add titles and effects and to color-correct.At the recent 2005 SMPTE Seminar on The New Digital Workflow, Rosenberg explained the method behind the madness. “The goals when I got involved, were to take the online through a process where we touched everything with a limited number of other applications and services. We wanted to keep it creatively powerful and financially tight. Here we had an environment where we were working on a DI in a space with creative tools that allowed us to do animated titles, scrolling titles and really good text work. We also did all of our dust and scratch removal—remember this was shot in primitive desert conditions with racing vehicles kicking up a lot of dust—in the same tools, so we didn’t have to pay for a separate pass to do dust and scratch removal.”One of the biggest controversies Rosenberg faced was color correction. He was determined to prove that he could color-correct in the Premiere Pro despite the prevailing attitude among post professionals that there is no viable color-correction tool for a PC desktop machine in a linear HD resolution. “We were able to achieve results so that even the folks at Laser Pacific said, ‘The film looks good,’” says Rosenberg. “We were pleased with that.”Working concurrently for nine months, Brown and Waugh shaped the story, sending project files back and forth between Long Beach, where Brown edited on an Avid Film Composer, and Hollywood, were Waugh edited on an Avid Express DV. As the picture was locked, Waugh output edit decision lists for Rosenberg at Laser Pacific, who started building reels on Adobe Premiere Pro in HD16:9. The footage was capture using BOXX RT system—CineForm Compressed 1080p (23.976). His main objective was to mix the differing formats and frame rates and bring the project up into an HD 10-bit 4:2:2 resolution, all of this with a desktop finish. Ultimately, the project needed to be output to film, D5 and Windows Media 9.The online assembly was standard. The HDCam content was downconverted to DVCam for the offline editing, retained the same timecode so when the 29.97 fps EDL was imported into the Premiere Pro and converted back to the original HD frame rate of 23.96 fps, the timecode lined up at every second despite the differing frame rates. The biggest hurdle was taking the 30i (DV) footage and converting it up to HD resolution, so that it looked as good as the more robust formats. Another positive attribute of this desktop workflow was that it allowed flexibility in the creative process. Titles and shots could be changed in the Premiere Pro up until the last minute.The lynchpin of the conversion process was a Terranex box, an essential component in unifying the formats and frame rates. QuBit drives were used to move the data around in a compressed 4:2:2 HD format. Rosenberg would export to the QuBit the final color corrected and dust-busted AE compositions that were rendered into single CineForm file for each reel. Laser Pacific would print to film with Arri Arrilaser printers. The QuBit files were also used to create the D5 master Exporting to Windows Media 9 for DVD mastering was a simple menu selection in Premiere.The specs of the system used for online included a BOXX Technologies HD-integrated workstation with a dual AMD Opteron, an external QuBit drive, a Xena HD/SDI—I/O with genlock and six-channel AES/EBU audio, 1.5 TB of Raid 5 storage and 3 GB DDR memory, a Sony HD monitor and an external waveform/vecterscope monitor. Rosenberg enthuses, “What I loved about our system was that we had one PC with 1.5 terabytes of storage inside, that’s up to 60 hours of compressed HD material. When you see the film projected, you’re going to see a film that’s projected from an HD DI. There are parts of the film where you see 16 or 35 from the source and you can’t tell the difference.”One of the things the team did to have a proof positive of their method was to project a film-out test. Everyone in town told Rosenberg that he could not work from a compressed file because it would have artifacts and look like a compressed jpeg file. In other words, it will look like “crap,” especially when color-corrected. Although, previous tests had been done up-rezing through the Terranex and printing to film, the CineForm DI codec that the project was using had not been used before, so Rosenberg insisted on further testing. He elaborates, “We took a mixture of all the footage that had been used in the film from bad Hi-8, bad mini-DV, real dynamic DV, high-contrast HD, real grainy Super 16 and we brought all those clips into our color correction tool and we pushed everything. We made extreme looks and then we made good looks and then we made normal looks. That way we knew, if we pushed color, do we see any anomalies. If we have color exactly where we want it, do we see any anomalies?”To compare the results, the team alternated between screening the digital print and the 35mm print. Rosenberg continues, “We sat in Laser Pacific’s digital theater and threw up a QuBit file—and I was really nervous at this point because everyone’s saying does this system work?“This is the first time we’re doing it, but I trust that it works, but we have to wait and see the results. The file appears. This is the digital file that we’re delivering to print to film. We start watching to see if anything sticks out that doesn’t look like the original. We passed the test.”The differences between the digital file and the 35mm print were so negligible that when the projectionist accidentally projected the film print, everyone thought it was the digital file. Rosenberg proved that a compressed file format could be output to film and still retain the essence of the original footage. The accuracy of the film-out gave the team confidence that they could take color where it needed to go.Rosenberg credits Laser Pacific engineering VP Doug Jaqua with keeping the look of the film print consistent with the look of the digital online. For Jaqua there is an inherent challenge when outputting to film from electronic media. The images are different. It is a matter of managing the densities and color between the two. As a color science engineer working with the differences in the two technologies, his goal, and that of Laser Pacific, is
to make the palette larger so as to accommodate the qualities of both media. “The strong point of what we achieved with this film post-wise is that acquisition formats are all in the eye of the creative person,” says Rosenberg. “If you want to shoot Super 16, you shoot Super 16. You don’t have to worry about your film print because you can online in HD and strike a print if you want to. The important thing is what it looks like when it is transferred to 35 for the film-out.”Dust to Glory postproduction is a case study from the digital edge. By pressing the limits of available desktop technology, the visionaries steering this film have demonstrated cost-effective independent filmmaking at its best.
Written by Mary Ann Skweres
