My work focuses on the intermediate stages of image generation in diffusion models, when images are still emerging and have not yet resolved. By stopping the denoising process very early, I work with images that are neither fully representational nor random. They have form and visual structure, but remain outside the kinds of images the model is optimized to produce.
My background as a theoretical physicist influences the way I approach these systems. Abstraction and the search for structure in unfamiliar spaces are part of both practices.
Earlier works translated transient computational states into painted grids, using the grid as both a visual structure and a reference to the underlying organization of digital images.
More recently, my work has expanded into language, translation, and interaction. Across these projects, I am interested in the unfamiliar spaces, images, and languages that emerge between human intention and machine process.