Turbulence
The piece explores ambient storytelling through abstract visuals.
Yet, the main view comes from the fact that all subjects and movements are envisioned by AI/ML tools. The piece exposes how algorithms perceive our world in motion, based on common archetypes but influenced by hidden, random factors that we can hardly anticipate. It praises the nature of generative methods and parallels it with the artwork’s topic, both being chaotic in a mathematical sense — inherently structured yet unpredictable.
Technical details:
The film was generated in one shot using a custom Stable Diffusion toolkit. Two kinds of inputs were fed into the model to achieve the desired effect: one included text prompts which drove the semantic narrative; the other was a set of images, ensuring a rich and consistent visual style all over the film. The motion and dynamics was devised by the AnimateDiff technique without time-based media guidance.
image: Vadim Epstein [eps696]
sound: Alexander Potekhin [Lazyfish] & Andrei Antonets [0ID]
code: Stable Diffusion, RAVE, NI Reaktor
2024