In Fungus Dreams, Valérie Bélanger explores the aesthetic possibilities of emerging AI tools by combining her own wildlife photography with the outputs of generative video system powered by machine learning tools. Bélanger uploaded still images, over and over again, each time extending the space of the photo into a increasingly dreamy landscape. The resulting video is a surreal machine dreaming landscape.
Fungus Dreams captures a sense of how many of us are feeling in 2024.
It undertakes a continuous cycle of eroding something picturesque, something beautiful, and transforming it something grotesque, but in the more Baroque sense of this word. In the dreamy colorful rounded landscapes, drawn by the machine learning system, there lies a monstrous element. Each of the morphing episodes moves away from a pictorial space with a kind of unpredictability that puts us on unstable ground. We are unsure were this hallucination will go, and our experience of machine hallucination, in this current time of AI super-development, leaves us somewhat wincing… waiting for the machine system to reflect elements of our darker human nature.
The commissioned version of Fungus Dreams is intermittently on display at Artengine.
Below is a short extract from Fungus Dreams.
Fungus Dreams
Valérie Bélanger
10 minute video loop (no sound)
This work uses the RunwayML platform with a particular focus on using images as prompts to create short video files.
The research for this work began in a studio course at the University of Ottawa exploring the artistic possibilities of new tools in machine learning and AI, and Artengine commissioned a full looping version for presentation in its facility. We worked with Bélanger to bring her experiments into a more complete space of a finished work.