The Association of Moving Image Archivists (AMIA) Digital Asset Symposium 2020 shows Digitization and Artificial Intelligence tools unlocking new potential for archives
The human impulse to create, and to save those creations for future generations, is at the heart of the archival profession. But how are these seemingly timeless impulses transformed by current and emerging technologies? The Association of Moving Image’s2020 Online Digital Asset Symposium (DAS) showcased a diverse set of case studies to explore developments in digitizing, analyzing audio-visual materials with artificial intelligence, and long-term storage. With an eye to the future of archives and how assets are stored and meaning is distilled from them, DAS 2020 was an energizing and inspiring peek into what lies ahead.
From Family Film to Museum Object
Although we have become accustomed to the ubiquity of mobile phones and their ability to record, edit and broadcast videos, it was not long ago that some of the only people who had access to cameras to film their family vacations were movie studio executives. This is one of the stories that was shared by the curators from the Museum of Modern Art’s Film Department presenting the case of Private Lives Public Spaces, a large-scale exhibition of amateur and home movies.
The curators described the pleasure of handling and screening the original films in a “gloriously hands-on object analogue process.” They also reflected on the complexity of curating, digitizing, and installing this unique collection in a gallery setting.
In a homage to MoMA’s historic photographic show Family of Man, the home movies were presented in a tightly organized collection of flat screens on the walls. Many of the screens had to be custom made in China so they would match the aspect ratio of the original film. The curators shared observations about how much has changed in our relationship to recording and sharing home movies and wondered how today’s home movies might be archived and accessed in the future.
Recording Artists and Dragons: Preserving Pop Culture
Iron Mountain Entertainment Services’ presentation The Art of Media Preservation & Digitization explored technical challenges and solutions for digital migration of magnetic audio and video. Sharing their experience digitizing the Tupac Shakur Estate’s Amaru archive to preservation-quality digital assets that will last up to 80 years, they focused on the importance of gathering and organizing metadata, new tools for Video Up-Resolution, and the challenges related to maintaining legacy playback equipment. Speaking to the shortfall in generational knowledge transfer, they expressed hope that younger people would become interested in yesterday’s tape technology so that migration services can continue transferring magnetic tape into the future.
Archivists from HBO presented their massive project to archive the physical objects used in the production of the eight seasons of Game of Thrones. The project began while the final season was wrapping, allowing the archivists to handle the management of the wrap, and bringing in the Assistant Propmaster, Costume Supervisor and Production Coordinator into the archival team.
By fall of 2018 there were 19 people in Belfast working on the archive, which used unique identifiers for 54 different types of assets including construction, wigs, fake cadavers, candles, and costumes. HBO’s motivation to maintain this highly detailed archive is to be able to create exhibitions, a possible fan experience in Belfast, and to support possible future prequels. The archivists explored how the corporate perspective on the financial value in the archive sits in contrast to a heritage-oriented approach where value is assigned to the cultural significance of an archive and not its profit potential.
The director of the Louis Armstrong House Museum shared their mass digitization project supported by a $2.7 million private grant made by the African American investor/philanthropist Robert F. Smith. The grant allowed the Museum to digitize and photograph over 60,000 items including 16,000 photographs and negatives, 356 videos, 500 musical arrangements, 111 scrapbooks, 315 posters, and 7 trumpets; thus creating the world’s largest archive for a jazz artist.
To complete this massive digitizing project the museum worked with the company Deluxe who helped to create a custom platform, met with metadata specialists, and found jazz history experts through Craigslist to create metadata. They used additional vendors for digitizing outside of their standard in-house workflows, including Mayer Media for obsolete audio formats, and 3rd Eye Digital to photograph every 3D object in the museum. All assets are available with a watermark on the museum’s website, and permission for licensing can be requested through the site as well.
The Museum’s collection is a reflection of Armstrong’s own strong archival impulse and the subsequent commitments made by the New York City Department of Cultural Affairs, the City University of New York, Queens College, and donor Robert F. Smith, who recognized the significance of this collection for documenting the life and impact of a seminal African American artist.
Archives and State of the Art Technologies
The previous cases raise a key question: how will we access and learn from all these important assets once they are digitized?
In the presentation Bursting the Inverse Bubble: Audio and Video in the Information Economy Chris Lacinak, President of AVP built a very clear case for the need for searchability of video assets. Giving illustrations about how important full-text search is in our lives, he showed how our current inability to search within video assets for specific information puts limits on the value and usefulness of these resources, especially at a time when the use of online video is exponentially increasing. Lacinak shared a variety of tools/platforms that do offer automated video search features but explained that because the platforms are closed, there is a lack of democratization and interoperability. The presentation highlighted AVP’s product, the Aviary Platform, which promises to “better leverage and manage robust metadata” through the integration of a variety of machine learning tools and standards.
The emergence of Artificial Intelligence and Machine Learning will not only give us the ability to search in new ways but also to revisit assets for entirely new purposes. Tape Ark presented a talk on Digital Data Preservation Across Industries that explored the potential for data re-use in the machine learning age. Seagate/Tape Ark is helping companies move vast amounts of content in local storage to the cloud, such as the data picked up by sensors that energy company vessels tow for hydrocarbon discovery. The migration of this data and the tools to analyze it could allow for the original purpose of the data to be flipped: data that was collected in the 1970s for oil and gas extraction could now be used to explore opportunities for carbon sequestration to address climate change.
The presentation Smart Stacking II: Machine Learning and Information Science at PBS explained how they are layering the services of the GrayMeta Curio platform and the semantic tools of the Pool Party company to delve more deeply into the information contained in video assets. With GrayMeta, PBS analyzed the children’s educational series Sesame Street – training computer models to identify and label Sesame Street monsters, and to transcribe children’s voices. Next, they are working with Pool Party to correlate taxonomies with the pre-school curriculum. These processes explore new ways that Artificial Learning and Machine Learning will be able to help educators and students learn from their rich archive in the future.
With so much data being created and stored, how can archivists plan to store data for the distant future? Microsoft Research’s presentation Project Silica: Glass – The Future of Long-Term Archival Storage begins with the premise that companies and organizations are producing more data than they are currently able to store because of the high costs associated with hard drive technology and LTO tape. Citing the lack of evolution in storage technology, Microsoft Research is developing a way to store compressed data inside of small chips of glass in order to transform the way that large data centres operate, bringing more data into the cloud for users.
Conference attendees wondered how users in the future would be able to decode data stored on glass. Notions of embedding some sort of visual “rosetta stone” onto the glass were explored, which would allow people in the future to understand the technology that might be needed to read the glass. Although Microsoft is developing proprietary technology to read the data embedded in 3 dimensional voxels, the presenter mentioned that microscopes could also read it, but not at speed.
An Eye to the Future of Archives
Although this technology is still in its infancy, it is an important reminder for archivists to think about the future-oriented nature of the archive in a time of drastic technical transition. Last year’s 2019 DAS included a presentation about storage on synthetic DNA, and archives such as The Netherlands Institute for Sound and Vision and The National Film and Sound Archive of Australia have both recently started exploring the use of this promising storage medium which may be accessible for 100,000 years.
How will future generations understand the world of today? Recognizing the responsibility of archivists to put forth a decolonial perspective, the NFSA and their partner on the project, the Olympic Foundation for Culture and Heritage (OFCH) encoded onto DNA the 49.11 second gold medal run of Cathy Freeman in September 2000, the first Aboriginal person to win Olympic gold in an individual event. Steps such as this, and the establishment of AMIA’s new Diversity and Inclusion Fellowship Program, illustrate the importance of archives becoming more self-reflective, and in the process allowing future generations access to the richness and complexity of today’s societies.
The Association of Moving Image Archivists Conference will take place online 17-20 November 2020.
Randi Cecchine is passionate about the potential for Artificial Intelligence in A/V archives. She is currently researching how institutions are implementing the use of AI systems. Would you like to collaborate in her research with a brief interview? Please contact her via email at firstname.lastname@example.org or via Twitter at @RCecchine