Making the DCP: Putting digital standards into practice at Deluxe Technicolor
For exhibitors and distributors, the challenge has been to make digital as universal as 35mm film, to make sure movies created anywhere in the world can run on projectors anywhere else. That took the energies, experiences and time of many talented people—and their efforts continue. More than a decade and a half after work on the standards began, various industry organizations—Digital Cinema Initiatives LLC, Inter-Society Digital Cinema Forum, the American Society of Cinematographers, the Society of Motion Picture and Television Engineers and others—continue to update the standards and revise the specifications.
But the requirements are just the starting point; Digital Cinema Distribution Masters and Digital Cinema Packages have to be made to conform to them. Today, much of that is handled by Deluxe Technicolor in Los Angeles, where Sean Romano, senior VP, global mastering operations and security, leads a team of people who do that work.
Film Journal International: The industry talks about Digital Cinema Distribution Masters and Digital Cinema Packages. What do those terms mean for you?
Sean Romano: We use Digital Cinema Distribution Master—the DCDM—as a catchall term for all source elements that go into the Digital Cinema Package. Those “masters” include color-corrected image files, mixed and equalized audio files and translated subtitle elements. There are a number of suppliers of those “masters” and 95 percent of the files we receive from them, we receive electronically—via standard transfer protocols over our 10-gig data pipe. On-site we have about 1PB—a little over one thousand TB—of storage, all highly secure.
FJI: How do you maintain security for the content?
SR: Everything on-site has been designed around strict MPAA security protocols. Although we don’t discuss our security parameters, we have multiple processes and procedures, as well as specialized hardening equipment. We’re mindful of the value of the content we’ve been entrusted with and we keep it fully protected throughout the process.
FJI: How far in advance of the movie’s release do you receive the elements?
SR: Generally, we receive the source elements anywhere from two to four weeks before release. Normally, image files come in reel-based—reel one followed by reel two, and so forth. Audio comes in separately and subtitle files follow—usually within a day or so after we receive the image and audio material. The image files and audio files we receive are uncompressed; we need to compress the image files.
FJI: Talk to me briefly about compression and decompression.
SR: For a 2K resolution feature, you’re looking at 1 to 2TB of uncompressed image data. A 4K feature has 6 to 8TB of uncompressed data. We have to make it manageable for the theatres, so we apply variable bit-rate compression, at up to a 10:1 ratio—meaning, for every second we can have up to 250MB of data. That takes either a 2K or 4K feature down to 200GB of compressed data—which puts it in a manageable range. At the theatre, the compressed JPEG2000 file that we create and distribute on the DCP is decompressed by the secure media block into a TIFF format and, using the keys, the image files—in that uncompressed format—are displayed through a projector.
FJI: Let’s talk about how you turn the image, audio and subtitle files into that DCP.
SR: We encode the image files—encrypting and compressing them—and wrapping them, which means we put all of the files together into a compressed MXF file. We encrypt and wrap the audio files—putting all of the channels together in an uncompressed audio MXF file. And then we package all the MXF files with any other “essence” files—subtitles, etc.—which will be the heart of the Original Version, or OV. All that work is done by our Mastering Operators.
FJI: So what else is on the OV?
SR: An OV—or any DCP—is a group of “essence files”—image, sound and subtitles—and “control files”—those that describe how the essence files will be used. So, in addition to the encrypted and wrapped picture and audio files and subtitles, the OV has a CPL—the composition playlist that lists the order the files play in and what picture files play with what audio files; and a PKL—the packaging list which lists all the files in the OV. The OV also contains an Asset map—which is a list of the names and unique ID numbers of all the files the OV contains.
FJI: How long does it take to make the Original Version?
SR: It depends on the feature. 3D OVs take longer to make than 2D. A three-hour movie takes longer to make than a 90-minute movie, etc. Generally, our Mastering Operators spend about two to four hours creating the original OV—and then an equal amount of time checking and evaluating it. In most cases, the OV is evaluated in our screening rooms that mimic the actual theatrical experience. We actually check movies twice: We watch them once to be sure our work is error-free; and then studio reps—the director, the editor, the cinematographer, or others who know the film really well—will watch them to be sure they’re fully satisfied. We have people in our screening rooms watching movies 24 hours a day, 365 days a year.
FJI: What adds complexity to making the OV?
SR: The variety of new technologies—including High Frame Rate and High Dynamic Range imaging, the multiple immersive audio systems, and projection systems—have changed the work we do. And, in some cases, highly compressed deadlines with multiple last-minute changes to comply with filmmaker requests and foreign country requirements bring added complications. But it’s all part of the job.
FJI: Are there multiple OVs for each movie?
SR: Currently, for the domestic market, we make anywhere from two OVs to as many as twelve for a particular movie. The number of versions depends on the complexity of the distribution need. For example: 3D features may have multiple OVs—one with a 4.5fL light target, one with a 7 ftL light target, one with a 14 fL light target. And there may be multiple audio versions—for 5.1, 7.1, Atmos, DTS-X, Auro and so forth. Internationally, there can be German OVs, French OVs and so forth—each with multiple versions.
FJI: Once the OVs have been created, how are they used to deliver DCPs to theatres?
SR: Elements of the Original Versions can be used to make variations of that element for the other DCPs. Making DCP copies starts with merging all of the essence files on the OV into one set of files—and then outputting that to a “media element.” We then replicate that media element—with different “instructions” for how those will be shown. The result is a set of files tied to the length and complexity of the feature. But with variable-frame-rate image compression, animated features usually compress to a small file size, while heavy action movies with lots of smoke, fire and open field scenes are often somewhat larger. The size of the DCP files will generally dictate how much storage space is necessary on a hard-drive deliverable.
FJI: Is it possible that a DCP contains more elements than are actually being played in the theatre?
SR: Yes. We generally try to include the most versions we can in a single deliverable, so there may be multiple audio files and a complete image file in 2D and 3D on the hard drive. CPLs—content playlists—determine what plays, where. They can even dictate seamless “censor cuts” for international versions. The keys only work with one CPL.
FJI: So how large are the DCP hard drives?
SR: The DCP is put onto a hard drive that can hold as much as 2TB or a small USB drive—for trailers—that can hold as little as 4GB. In the U.S., we generally use 1TB SATA hard drives inside of a CRU data-port enclosure. For international territories, we send drives ranging from 250GB to 1TB. In addition to terrestrial delivery, we also use satellites; because the data underlying the DCP is file-based, the same files sent via hard drive can be transmitted electronically. For domestic distribution of the tentpole movies, we generally make anywhere between 2,000 and 3,000 hard drives—and as many as 2,000 additional theatres may receive the movie via satellite.
FJI: What happens after the movie’s run ends?
SR:If the files were delivered via satellite, they’re simply deleted from the theatrical servers. If the files were delivered on hard media, the drive is returned and put back into available inventory. It’s then reformatted and reused. We get about four to six uses from each hard drive per year—with a lifecycle of three to five years—so our rule of thumb is that a hard drive can be erased and reused up to thirty times without a problem. The majority of our hard-drive replication and distribution takes place in Wilmington, Ohio.
FJI: How many versions do you make for a typical big-budget movie that will have international distribution?
SR: Among our different offices, we make DCPs for the world, so for a 2D-only feature, we might make 60-plus versions; if the movie will be distributed in 2D and 3D, we might make 300-plus versions. For some theatrical releases, we’ve delivered upwards of 600-plus different versions. We make DCPs for all audio and image formats—and for about 85 different languages across the world. Those versions get their own “sign-off”—which differ client by client. We’ve also made trailers that have had over a hundred versions—including subtitled or dubbed versions for other countries.
FJI: How is trailer work different from feature distribution?
SR: A feature takes hours to package; a trailer can take fifteen minutes. While we may have several weeks to work on a feature, a trailer usually has to be made within three to five days of release. And of course, trailers are unencrypted; feature content is encrypted with military-spec cryptographic file protection—128-bit symmetric keys per track file.
FJI: Do you create keys?
SR: We create KDMs—key delivery messages—using information supplied by equipment manufacturers—the “public certificates”—client timelines, and a proprietary TDL—trusted device list—of equipment with “private certificates.” Each KDM is specific to an exact piece of content, a precise system in a designated auditorium, and an explicit time window set by the client. That’s been the process from the beginning; it hasn’t changed and because it works well, I don’t expect it will.
FJI: As you look into the future, what changes do you see in your part of the business?
SR: Digital standards have been evolving, so the industry has been aiming at a moving target. Now that SMPTE has finalized their specifications and guidelines, the industry is moving into a more stable phase. There will be further iterations with the introduction of new technologies, but “would,” “should” and “could” are being replaced by “must” and “will”—so content creators and equipment manufacturers are clearer on what’s needed. For us, I’m expecting that as systems become smarter and talk to each other, our work will become more automated with less need to create as many versions. Once upon a time, film was a very limiting system, but it had a simple elegance. Digital brought more choices and more capabilities—but also more complexity. I’m hoping that, as the “final” standards are adopted, the industry will find a way to combine the power of digital with the simplicity of film—and that will open new possibilities for everyone, including us. That’s what I’m looking forward to.