The project explores generative AI and its Surrealist foundations from a painter’s perspective. Its aim is to understand how these systems work, to demystify them, and to claim them as a space for artistic practice.
„Fluidity of Images“ is an artistic research of open-source generative AI models and workflows through original painting series: 30 animal studies in oil and acrylic, and 35 figure-and-landscape works in colour pencil and ink. These paintings become the sole material fed into a generative AI system trained to mix images using flux.redux model, producing hundreds of mixed images from pairings across and within the two series. Furthermore, a low rank adaptation model (LoRa for Stable Diffusion 1.5) is trained with mixed paintings for live performative interpretation of body images.
Video documentation from performative media installation, 2026, Stadtgalerie Saarbruecken
Each pairing of paintings/images opens what the project calls juxtaposition space: the navigable territory of possible images that exists between any two source works, where the two paintings or the body image from the camera can be combined at different strengths and balances to reveal a whole spectrum of outcomes rather than one fixed result. This continuous, in-between quality of the images connects the project to surrealism’s longstanding interest in juxtaposition, the idea that placing two unrelated realities side by side can generate meaning and imagery neither one holds alone, here extended from a technique of montage into an entire generative space to explore.
A further phase develops reflective AI as method: selected AI outputs are hand-redrawn and repainted, so that drawing becomes a way of learning from and returning to one’s own visual language, using the machine’s responses as a mirror that deepens the artist’s understanding of their own work.
The cycle continues with a large-scale performative media installation (video above). It captures visitors’ bodies with a camera and interprets them as animal paintings, which then flow into the picture frames on the projection wall.
The exhibition “Flüssige Körper” (Fluid Bodies, 24 October 2026 to 28 February 2027), explores Surrealist juxtaposition with generative AI. The show presents three intermediate stages of this process in three works: 21 prints of mixed animal images presented in light-boxes, 21 watercolor studies of these mixes, and a performative installation.
The exhibition is part of the Stadtgalerie Saarbrücken’s series “AI Joins Us”, which invites artists and audiences to reflect critically on the possibilities and effects of artificial intelligence. The concept for the series was developed by Hojin Kang, who was also the first artist in the series, and Katharina Ritter.
In the following more images documenting the process
