Skip to content

Glacier Trilogy – Part 1: a synthetic archive

#algorithmic #dark #engaging #experimental #hypnotic #immersive #thought-provoking

Materials and media

large scale video projection





Cittadellarte - Fondazione Pistoletto July 1 - September 25 2022 Faces of Water - BOZAR, Museum of Fine Arts Brussels, Oct- Dec 2022 G-STIC Rio - February 2023

Funded by


Residency at

Cittadellarte - Fondazione Pistoletto (IT)

Curator’s statement by Sofie Crabé (curator of STARTS4water exhibition at BOZAR, Brussels)

“Theresa Schubert developed a compelling, artificial intelligence-based audiovisual video projection that displays slowly emerging and dissolving synthetic mountain landscapes with glaciers… The collected data nourished three machine learning models. The machine learning algorithms transformed the data into an open-ended aesthetic collage. They assigned visual shapes to the meanings of words, revealing the glaciers in a new poetic way, while providing critical insight into the condition of the glaciers.”

The Glacier Trilogy is an immersive artwork investigating glaciers as the starting point of fluvial systems and the future of water in climate crisis. Produced by combining advanced computational technologies (such as generative adversarial networks, atmospheric sensors or real-time ice-fluid simulation for 8K) with sculpture materials and human creativity, the Glacier Trilogy (GT) stimulates an emotional engagement of its audience.

The work complex is based on Schubert’s research of fluvial systems in the Piemont area in North-West Italy. Schubert makes works that discuss the human impact on the environment, but also how we can use technology for better and more respectful relationships with nature, which is necessary as humanity is confronted with climate crisis. Glaciers hold an extreme importance not only as storages of water but also as a memory of the earths past and as indicators of climate change. Glacier ice archives millennia-old (an)organic information, such as (micro)organisms, pollen, and atmospheric dust, allowing scientists to acquire knowledge about ancient ecosystems and to predict future climate scenarios.

Re-imagining glaciers through AI or a synthetic archive of glaciers

This work is a video projection made by using a combination of machine learning models that shows the formation and abstraction of synthetic glaciers. Starting from abstract imagery, slowly mountain landscapes with glaciers start to emerge. They are based on an earlier training process with historical images of the Italian Glaciological Committee in Turin and the Sella Foundation in Biella in combination with text input from the Italian geologist Arduito Desio. Together with the visuals I created a dense sound composition consisting of field recordings, synthetic sounds and voice recordings done with an Opera singer. In the arrangement I layered the voice, sometimes as echos of each other, sometimes in a canon creating a doleful atmosphere.  What emerges are synthetic landscapes that don’t exist, a kind of machine dream from the future that tries to imagine what glaciers looked like in former times.

Rebuilding relationships with river systems presents the final works of artists Theresa Schubert and Joshua G. Stein, the result of their residencies within the project S+T+ARTS4Water (2021-2022), part of S+T+ARTS (Science, Technology and the Arts), an initiative of the European Commission promoted within the framework of the research and innovation programme Horizon 2020, supporting collaborations between artists, scientists, engineers and researchers.

Find out more about my residency and the development process on the STARTS4Water Journal.

Artist: Theresa Schubert
Colab implementation: Moises Horta Valenzuela based on StyleGAN, ESREGAN, VQGAN+CLIP
Voice & Singing: Joseph Schnurr

Physical resources:
Photo Archives of the Italian Glaciological Committee Turin and Sella Foundation Biella