Samantha Owens, Cass Fino-Radin, and Daniel Mauro
Electronic Media Review, Volume Eight: 2023-2024
ABSTRACT
For more than 50 years, American artist Gary Hill has pushed the boundaries of moving image art, language, and technology. An early innovator of video art and interactive computer-based installations, Hill’s work has been foundational to the development of intermedia and expanded cinema. This presentation focuses on the recent conservation of Hill’s Tall Ships (1992), a 16-channel interactive video installation built upon the combination of custom analog and digital technologies, now migrated to newer technologies while maintaining the unique imaging system at the core of the work.
Introduction
Tall Ships was originally commissioned for the exhibition documenta IX in Kassel, Germany, in 1992 and would go on to tour the world throughout the following decade. Now in the collection of Glenstone Museum (note 1) in Potomac, Maryland, Tall Ships consists of 16 black-and-white cathode ray tube (CRT) projectors custom built by the artist that hang in a long, completely dark corridor. In its original technological iteration, each projector connects to one of 16 LaserDisc players controlled by a DOS-based computer with a custom board and software coded by multimedia designer and frequent collaborator Dave Jones. Appearing in each projection is the image of a person standing or sitting in the distance. When visitors walk into the corridor, the only light emitted comes from these faint figures. The images are high contrast, yet soft with an ethereal quality, as there are no borders framing the projections. As a visitor walks closer to the image, their steps trigger electronic switches in the floor that then signal the figures in the video to get up and walk toward the visitor, meeting them face-to-face. The figures stand there, life size, wavering, staring back at the visitor until the visitor steps away, triggering the figure to also step away and walk back to the distant position where they initially sat.
Glenstone worked closely with artist Gary Hill, based in Seattle, Washington, and Small Data Industries, based in Rochester, New York, to conserve the work for future preservation and exhibition. The project is greatly indebted to Hill, not only for creating the unique artistic environment of the work and pushing the boundaries of contemporary art, but also dedicating his time to this conservation project and adapting the technology to its next iteration. The work of the artist and the project team ultimately makes the work more readily available to audiences to experience the singular and wholly original feel of Tall Ships. Necessary to its original creation and now its conservation has been tinkering with technologies in unexpected ways. The work’s 16 CRT projectors were found to be custom constructions of the artist’s invention, and their hand-made nature created a highly unique image quality. The CRT monitors used in the original projectors are Sony PVM 411, a black-and-white model with a four-inch screen, which retailed in a rack-mount array of four monitors. When building Tall Ships, the artist removed these monitors from their original rack mounting. Then, through various additions of aluminum fittings to the enclosure of each CRT monitor, he created an attachment system for a lens that was sourced from a “surplus store,” turning these small televisions into projectors.
Inspection and testing of these projectors found that numerous aspects of their hand-made nature resulted in the projected images’ distinct and striking quality (fig. 1). While the lens selected by the artist had a built-in focus ring and the lens mounting system allowed for additional physical adjustments of the distance and angle of the lens in relation to the CRT monitor, it was found that only one portion of the image could ever be in perfect focus. The focus then decreases radiating out from this point. The lens also introduced a distinct vignetting, with the center of the image area being the brightest. Finally, the overall projected image had virtually no visible pixelation. Seemingly due to refraction of light occurring in the interaction between the CRT and lens, the black-and-white image had a soft, glowing, slightly smeared quality. All of these factors combined to create an image quality and experience of viewing the work that feels otherworldly. As it was possible to source backup CRT monitors and to service and maintain these projectors for the foreseeable future, the decision was made not to replace them, but instead build the capability to care for them as they are.
Another aspect of the installation left unchanged are the original sensors used by the artist to detect a viewer’s presence (fig. 2). The in-floor, pressure-sensitive switch mats were found to be consumables that eventually wear out with use. However, at the time of treatment, the manufacturer was still in operation and producing the exact same model of sensor with no modifications. As such, backups and replacements were acquired.
Therefore, our treatment focused on replacing the following behind-the-scenes equipment: the original PC, the LaserDisc players, and the software engineered by Hill’s longtime collaborator, Dave Jones. These three components combined to form the beating heart of the installation, dictating what is shown on the projectors, the logic and flow of its interactivity, and its “feel” to the extent that any playback equipment that involves interactivity always imposes its inherent limitations due to response times and latency. Before assessing viable contemporary technologies to replace this original stack of equipment, a thorough analysis and understanding of the original equipment and software’s behavior was critical. This was achieved through numerous conversations and interviews with Hill and Jones, inspection and analysis of the work’s original C++ source code conducted by technician Emma Dickson, direct observation of the original equipment in operation, and, finally, rigorous instrumented observation and analysis of the equipment’s behavior for reverse engineering (fig. 3). The depth at which we could understand the subtle behaviors of the work in its original iteration was only possible due to two very fortunate factors that are never guaranteed. First, Jones was able to find the source code in his archives and share it for analysis. Second, the original equipment functioned properly. Our inspection and analysis would have been notably incomplete and inaccurate without these two aspects.
Our analysis found that the work’s interactivity relied heavily upon frame numbers that were inherent to the original LaserDiscs. The various sequences of the characters depicted in the projections walking, sitting, and standing were defined as different clips within the source code using start and end frame numbers. Numerous aspects of interactivity were dependent on querying the current frame number of the given disc being controlled. Finding that the original LaserDisc players could overlay the image’s current frame number, we set out to create highly detailed documentation of the work’s behavior and timing related to interactivity, frame sequences, and response times. To conduct this, we devised a testing setup that included showing the work on an LCD screen with analog input and a custom Arduino circuit that simulated the switch mat sensor using an electronic relay. This circuit had a button to initiate the transition from open to closed circuit and provided indicator lights—red for open, yellow for imminent change state, green for closed—and an alert sound that beeped the instant the circuit closed.
Mounted above this testing apparatus was an iPhone 13 Pro Max recording video at 60 fps. For each installation channel, multiple takes of footage of the testing apparatus were captured for every possible interaction state. The subsequent footage provided us with numerous capabilities. As the LaserDisc frame number was overlaid, we were able to cross-reference the observed frame numbers with the source code supplied by Jones (fig. 4). This confirmed that the version of the source code provided was identical to the compiled version we were observing. We were also able to verify that there was no variation or deviation between the source code’s instructions and the real-world behavior of the equipment. It was precise to the frame number perfectly each time. Crucially, the footage we captured gave us the means to measure any potential latency in response times.
Raspberry Pi was selected as the hardware platform that the new iteration would use because of its analog output with GPIO pins—critical for the direct sensor connection—as well as its durability with no moving parts, and its Linux core, which ensures long-term support. Each video channel was assigned a dedicated Raspberry Pi, and an additional unit was dedicated as a central control unit responsible for monitoring the overall state of the system, issuing overrides as needed for unique behaviors. For the main programmatic logic and control of the new iteration, we selected Python as the programming language due to its ubiquity, uncompiled and human-readable nature, and proven popularity in conservation and digital preservation. For video playback, a thorough assessment was conducted of pre-existing tools for Raspberry Pi that could be controlled via code. Using something off the shelf, such as VLC, would not provide the level of granular control and state monitoring that the work requires. As the new iteration’s code (co-written by Emma Dickson and Cass Fino-Radin) handled all video playback directly within Python by way of the Open CV module, enabling powerful video playback and manipulation capabilities.
All the Raspberry Pi computers were housed within a rugged rack, which allowed for permanently and internally wiring all equipment and designing a plug-and-play solution for the new iteration. All Raspberry Pi computers were connected via a network switch within this rack. To facilitate communication between the units, we selected the MQTT protocol for its robustness and resilience in the face of latency or errors. During the building and software development process, returning to the reverse engineering footage of the original iteration was found to be crucial for assessing the accuracy of timing. The original LaserDiscs were meticulously digitized to build this new iteration in digital form. When this digitization was completed, two sets of material were created: one with and one without frame numbers burned in. This allowed us to optionally display frame numbers when testing the new iteration and employ a testing workflow identical to how the original iteration had been tested. As a result, we were able to take footage of identical interaction sequences, set them to a synchronized start point using a digital video editing program, and ultimately assess timing fidelity to the original iteration on a frame-by-frame level. By measuring latency or timing, as our source footage was shot at 60 fps, this method provided accuracy down to the closest of 0.0167 second (or 16.67 ms).
After installing and observing the new iteration in person for the first time with Hill, this level of accuracy and analysis proved invaluable, as the decision was made to fully replicate the latency the original LaserDisc players exhibited when skipping between clips. Throughout the development, testing, and refinement process of creating the new iteration of Tall Ships, there were numerous instances where the new iteration’s behavior seemed perfect to the human eye. Still, upon closer inspection using the preceding methodologies, we identified and rectified subtle inconsistencies and inaccuracies that otherwise would have gone unnoticed. This finding points to a need for broader use of measurable and quantifiable methods of assessing the fidelity of treatments in the conservation of time-based media.
While both Hill and Jones were thoroughly involved before and during our treatment process, much of their input was necessarily technical (fig. 5). To complete our treatment, we spent focused time with Hill discussing and documenting the artwork’s more holistic experience and identity. Time was spent at Glenstone working in collaboration with Hill to look at, interact with, and test the new iteration of the work, which had been installed in partial form. The artist’s feedback resulted in a deeper understanding of the work, proving critical for its long-term stewardship. Numerous aspects of setup and calibration were revealed to us that were previously undocumented and likely lived solely as muscle memory within Hill. A formal artist interview at Glenstone, after spending two days collaboratively installing the work with Hill, further enriched our holistic documentation of Tall Ships.
An additional positive outcome of this conservation treatment was the significant reduction of energy usage of the work due to lower energy demands of the new equipment (fig. 6). A significant decrease was also made in the artwork’s carbon footprint for storage and shipment due to the diminutive size of the latest equipment. Through directly measured energy usage, we found that the energy consumption of Tall Ships has been reduced by 71%, and through carbon footprint calculations, it is estimated that the carbon footprint for a three-month loan from Glenstone to a theoretical exhibition in Valencia, Spain, would be reduced by 55%.
Much of the success of this significant undertaking was only possible with contributions from both the artist and technician. Without Jones’s source code and the knowledge of the work embodied by both Jones and Hill, our treatment would have been exorbitantly more challenging and costly. It was fortunate that this project was undertaken while the playback equipment was still functional, due to its unique and irreplaceable nature. Had it been found that the original equipment was not functional, our efforts would have been rendered even more difficult, and perhaps impossible. These dependencies point to successful conservation treatment being a matter of timing. While exhibition schedules often drive the prioritization of conservation work, this project serves as a compelling example when the care and migration of a particular work of art must be triaged so as not to miss a very transient window of opportunity. The conservation treatment of Tall Ships has illuminated the unique challenges and approaches inherent to working with custom analog and digital moving image technologies and best practices for preparing them for a fast-moving future of ever-evolving time-based media art.
NOTE
- Edition 1 of 2.
AUTHORS
CASS FINO-RADIN
Founder and Lead Conservator
Small Data Industries
Rochester, NY
cass@smalldata.industries
DANIEL MAURO
Assistant Curator
Glenstone Museum
Potomac, MD
daniel.mauro@glenstone.org
SAMANTHA OWENS
Associate Conservator
Glenstone Museum
Potomac, MD
samantha.owens@glenstone.org