Cypress Trees: Epochs

This series of photographic prints is composed of generative still works extracted from multiple points within the training runs of a GAN trained on the cypress dataset. Rather than presenting a single resolved output, these images capture moments from within the generative process itself, revealing the system as it learns, misinterprets, and reorganises its source material. Across the series, trees, sea, and sky appear as fragmented and shifting forms that hover between documentation and abstraction, where blocks of colour sit alongside moments of unexpected detail. At first glance, the images may read as landscape photographs; on closer inspection, their strangeness becomes apparent. Forms fracture, textures repeat, and spatial relationships subtly collapse, exposing the underlying logic of the machine’s vision. The works hold together two modes of seeing at once: the recognisable elements of a threatened environment and the synthetic distortions produced through algorithmic interpretation. The series foregrounds process over resolution, allowing the instability of the generative system to remain visible as an integral part of the image.