About Psionic Studio
PsionicStudio is a solo practice built around the intersection of machine perception and human intent. The work begins with a question: what happens when you give a neural network a feeling rather than an instruction?
Every piece is a negotiation — between the latent space of a model trained on centuries of imagery and the specific visual instinct of the person directing it. The results are neither purely algorithmic nor purely hand-made. They occupy a third territory.
THE TOOLS
The practice runs on a combination of diffusion models, neural style transfer networks, and custom latent-space navigation scripts. Stable Diffusion, ControlNet, and bespoke Python pipelines form the core — but the tool is never the point.
What matters is the feedback loop: generate, observe, redirect. The machine proposes; the eye decides. Most sessions produce hundreds of frames before a single image earns its place.
THE PROCESS
Each series begins with a visual hypothesis — a mood, a tension, a specific quality of light that doesn't exist in any photograph. The process is then one of approximation: training the model toward that feeling through prompt engineering, fine-tuning, and iterative selection.
Style transfer layers add a second dimension — the texture and tonal logic of one image applied to the structure of another. The results are often surprising. That surprise is the work.
The studio takes on a small number of commissions each year — editorial, fine art print, and experimental digital projects. If the work resonates, reach out.