Matt Gingold & Philippe Pasquier
Frame Biennial of Dance
Melbourne ― Australia ― 2023
Conference for Human Interaction (CHI’ 2017)
Denver ― United States ― 2017
Zentrum für Kunst und Medien
Karlsruhe ― Germany ― 2016
One Art Space
New York ― United States ― 2016
International Symposium of Interactive Art
Vancouver ― Canada ― 2015
Surrey Urban Screen
Vancouver ― Canada ― 2013
Longing + Forgetting explores the ways in which machines are ascribed intelligence, whilst humans are increasingly treated like machines. Combining physical and algorithmic choreography with architectural projection, the work questions the relationship between the inanimate and the animate: how are our bodies measured and controlled by the structures of technology? And what traces do our bodies leave on the structures around us?
Dance on film – indeed film itself – is deeply tied to the origins of ‘capturing’ and measuring human movement. As a precursor to motion pictures, the chronophotography of Étienne-Jules Marey, and more famously it’s use by Eadweard Muybridge to capture the walk cycle of people and animals, were entirely motivated by measuring the body and comparing the human with the non-human.
Film also offers the possibility of extending dance and augmenting human movement, as witnessed in the layering and echo of bodies in Norman McLaren’s Pas de deux , and the algorithmic re-animation of the walk cycle by physically rotating projectors in Dumb Type’s media installation work, Lovers (1994).
Longing + Forgetting [LAF] combines these histories of optical control and filmic expansion of the body, extending it to the architectural scale. Both critiquing and using machine learning, we wanted to create a work that animated architectural dance through projection technologies, but also evoked an immediate, emotional and embodied response.
Here the walk cycles of Muybridge are extended to other modes of traversing space: crouching, crawling, inching, swinging across the facades of buildings. As bodies are doubled, tripled, quadrupled, endlessly searching for something that is never really found, these ghostly apparitions re-animate the facades they are projected onto.
Their jerky movements, bounding boxes, repetitive actions, aerie stares – and the motif of hugging empty space and falling backwards – are both representative of the technologies increasingly used to surveil our bodies and lives, and the underlying technologies used to algorithmically choreograph the performance itself. The sound is also modulated and triggered by these spatial movements, with a sonic pallet of sub-vocalisations – moans, groans, breathe – mixed with time-stretched, computer-generated phenomes that evoke sci-fi engine rooms; epic spaces filled with the sounds of human efforts.
Longing + Forgetting is a collaboration between media artist Matt Gingold, Philippe Pasquier, a world leading expert in machine learning and director of the Metacreation Lab, and Thecla Schiphort, a pioneer of digital choreography who co-created Lifeforms, the choreographic software used by Merce Cunningham.
Commissioned as an interactive public artwork for the Surrey Urban Screen (Vancouver, 2013), the work was developed during a residency at the School of Interactive Art and Technology, Simon Fraser University (Vancouver, CA) as part of Moving Stories, a broader research project into gesture, machine learning, dance, movement and meaning.
It has had several presentations, including live performance, generative video projection (Generations, ISEA Vancouver, 2015), real-time data visualization (Scores + Traces, New York, 2016) and an academic paper and performance for CHI 2017 (Denver), the premier international conference in the field of Human-Computer Interaction.
A decade after its original creation, LAF was reprogrammed, remastered and presented by the Centre for Projection Art for the first time as an indoor, intimate, life size projection in the “unrestored zone” of the South Magdalen Laundry at the Abbotsford Convent for Frame: A Biennial of Dance in Melbourne 2023.
CREDITS
Concept
Matt Gingold, Philippe Pasquier & Thecla Schiphorst
Video Design & Code
Matt Gingold
Sound Design
Philippe Pasquier
Choreography
Thecla Schiphorst & Matt Gingold
Editing & Animation
Josh Burns & Matt Gingold
Lighting Design
Ben Rogalsky
Set Design
Greg Snider
Performers
Shannon Cuykendall, Matt Duncan, Sarah Fdili Alaouim, Meghan Goodman, Marcus Marshall, Joshua Ongcol, Priya Rajaratnam, Bladimir Santos Laffita, Nathalie Sanz, Cara Siu, Yawen Wang, Martin Wong
Photography
Matt Gingold & Anatol Pitt
MATERIALS
Variable (as presented at Frame Biennial of Dance)
1 x Mac Mini M2 Pro, 1 x RME Fireface UCX, 4 x HD 12000 ANSI L1505 Epson Projectors with ELPLU03S UltraWide Lenses, 1 x Sollinger LaserAnimation PHAENON Accurate 7.6W RGB Laser, 4 x 300W Full Range Monitors + 600W Sub
DIMENSIONS
Variable (depending on venue and façade)
9000mm W x 12000mm L, x 5000mm H (Frame Biennial of Dance)
Minimum 6000mm x 2000mm (using 2 x projectors)
HOW IT WAS MADE
Intentionally, the human labour highlighted in the creation and presentation of LAF is exactly what is erased in the marketing and debates around AI Art and machine learning in general. Whilst remarkable – even beautiful – these technologies do not occur without significant human intervention and environmental impacts, eg., squalid rare earth mining conditions; compute power, heat and water usage; the low or unpaid labour of data classification and the scraping of images and text without attribution, consent or payment to the original creators.
At the beginning of the project, we engaged in a collaborative rehearsal process with 12 performers to imbue otherwise everyday movements with an intensity of feeling. While the ‘movement alphabet’ is simple – walking, turning, crouching, crawling, climbing, jumping – it was vital that these movements be more than ‘mechanical’ motions, but somehow speak to the intimate stories and emotional worlds technology often occludes.
Using a custom built 5m tall climbing wall, we then filmed the performers as they interpreted each of the movements with different intentions and intensities. This resulted in a large dataset reflecting the diversity of bodies, cultures, movements and affect (or in Laban terminology, ‘efforts’) of each performer. In total there is over 4 hours of final footage across 2000 individual clips hand annotated by movement type.
In order to further prepare the video for ‘algorithmic choreography’, we then had to painstakingly hand animated each video clip. For example, if a performer was walking right, we needed to ‘keyframe’ or move the video to the left, with just the right amount of ‘energy’ easing in and out to sell the illusion of the performer walking on the spot. This process created a corpus of imperfect movement loops, resulting in the glitchy jumps one observes in the installation.
Finally, we developed a path-finding algorithm to ‘stitch together’ generative combinations from those movements, allowing us to map architectural facades by marking out spaces where performers can stand, climb, walk, crawl and sit.
As such, each digital performer is comprised of many fragments of movement – both handmade and machine processed – which can be combined to create complex and emergent choreographies.