Two of my favorite moments at work are when I confirm that the regular UPS visit just picked up a shipment of tapes going to the vendor and when I receive a package from the vendor—the hard drive with files from tapes sent previously. Yes, I send tapes, and in a couple of weeks I get files. Several emails and phone calls are usually involved between these two exciting moments. Digitizing tapes through outsourcing is my daily responsibility.
I am the moving image specialist in the Digital Conversion Unit at the University of Michigan Library. My project is to get moving image materials in the Special Collections Library digitized for preservation and access. The Special Collections Library archives hold a total of around 4,500 moving image items—and that doesn’t count materials in other
libraries on campus or in the University Archives. Among the twenty-seven collections with moving image materials in the Special Collections holdings, the Robert Altman Collection is the largest and the most important. From recorded interviews with Altman on various occasions to all the dailies from his last two films, A Prairie Home Companion and The Company, this archive is the only exclusive collection of documents from the life and career of one of the greatest American independent directors.
I was hired in 2011 to launch a moving image digitization program at the University of Michigan Library. The year before an audio specialist had initiated an audio digitization program: she researched digitization specifications and metadata, wrote an RFP and selected preferred vendors, and designed the workflow for vended digitization of rare/unique audio recordings in the library. I had been doing similar work for moving image when, in March 2012, a researcher for a Robert Altman documentary project requested access to over 150 recordings, with a strict deadline.
Thus, our initial test project quickly turned into an initial real project. The draft RFP became a Statement of Work for a specific list of recordings. Extensive and detailed communication ensued with the vendor. As many emails that went out to the vendor, almost same number were
needed internally to figure out where and how to place the huge files that the project would return.
Our basic goal—to create 10-bit uncompressed files as archival masters—was based on the assumption that all items in the collection are unique and rare. A question arose when digitizing non-unique items, such as VHS recordings of The Tonight Show where Altman was interviewed. The NBC Archive preserves the entire show and the University Library does not have any right to license such items anyway. Do we still need to create enormous files from such content? No. All we need would be an in-library access file with decent quality. With this simple answer, a schedule of preservation levels for moving images was prepared—a list of alternate digitization strategies based on the nature and content of the original recording. All exceptions to making 10-bit uncompressed files are listed with controlled rationales, which will be recorded in the metadata.
While this strategy appropriately aims to minimize risks of loss, the resulting file sizes are still enormous. We calculated that the 4,500 moving objects currently in the Special Collections Library would result in some 250TB worth of files. The mechanics and potential costs of accommodating these large files, even if they are generated incrementally over a number of years, took library administration and the digital asset management team by surprise. Moreover, while the general trend at Michigan has been to push digitized collections into the HathiTrust digital repository for long-term management, HT does not yet have the infrastructure—programming, standards, or cost models—to handle audiovisual files. As a temporary measure we decided to purchase disk space from the University’s Information and Technology Services while we look for a permanent solution.
Storage was not the only issue. Developing the metadata schema has given us the most headaches. For the audio project, the University has been using the Audio Engineering Society (AES) schema with PREMIS in a METS document. However, as many archivists in this field are aware, metadata for moving image is a moving target. Many hours were spent on exploratory research and reaching out to other programs, but the efforts did not yield a clear-cut solution. And there was no time left to wait. After investigating and mapping several available options (reVTMD from the National Archives and Records Administration, SMPTE, PBCore, PREMIS, etc.), we chose a temporary solution—metadata in a spreadsheet—until we can find a better and more stable option. We tried to capture in consistent manner all metadata that is considered important by most professionals in the field, for upload into the schema we select later.
Am I announcing in public that the University of Michigan just solved its two biggest moving image digitization issues by resorting to “temporary” solutions? Yes and no. Our goal is to save and make available unique and rare materials which are deteriorating and otherwise can’t be used—and which researchers want to use now. We just couldn’t wait until everything becomes perfect and stable. We made calculated compromises that have let us move forward with our program in the real world, but should also permit us to preserve these files for the long term.
// JungYun Oh, Moving Image Specialist, University of Michigan Library