About This Project

What Was Lost

Three A3 sketchbooks. Charcoal on cartridge paper. Decades ago, they existed—rough, impressionistic stage designs for plays that had taken up residence in the imagination and refused to leave. Tennessee Williams. Federico García Lorca. Henrik Ibsen. Bertolt Brecht. Samuel Beckett. Designs for impossible productions, drawn in the fevered hours of early adulthood when the boundary between vision and reality felt negotiable.

Those sketchbooks were once submitted as a portfolio—evidence of a particular way of seeing that earned entry to formal theatre training. Then they disappeared, as things do. Moved too many times. Packed in boxes that never arrived. The paper is gone. The drawings are gone.

But the visions never left.

What Remained

Memory is not faithful, but it is persistent. Fragments survived: the silhouette of a horse suspended in a paper forest. Glass animals watching from wire enclosures. A greenhouse built from thorned rose stems, its glass fogging with decades of unlived breath. These images would surface unbidden—in half-sleep, in the dark of theatres, in the moment before the house lights dimmed.

Thirty years of carrying other people's stories across stages, and these designs remained stubbornly present. Not quite complete, not quite forgotten. Waiting for a form of resurrection that hadn't been invented yet.

The Method

Each design follows the same process. Begin with the text—not the plot, but the emotional architecture. What is the play really about? What does it feel like to inhabit its world? Then search for the fragments: what images survived the decades? What kept returning despite every attempt at forgetting?

The memories, it turns out, are remarkably vivid. The suspended horse. The wire enclosures. The thorned architecture of the greenhouse. These visions have remained sharp across thirty years—not faded impressions but persistent images, carried intact through every relocation and reinvention. The difficulty was never remembering. The difficulty was translation.

Describing a vision of imagination in words is an excruciating art. The gap between what the mind sees and what language can convey proves almost unbridgeable. And then there is the machine itself—trained on thousands of examples, gravitating always toward the familiar, the average, the composite of everything it has learned. Each prompt becomes a negotiation: pushing against the desire of training data to reproduce what it knows, coaxing it instead toward something it has never quite seen before.

The results are not reproductions. They are new objects made from old ghosts. Theatrical designs for productions that will never be mounted, stages that exist only in the space between memory and machine learning.

On Stolen Things and Excavated Ones

The discourse around AI image generation circles endlessly around theft. Stolen styles. Scraped artworks. The labour of human creators fed into machines that regurgitate approximations without consent or compensation. These concerns are legitimate, and this project offers no resolution to them.

But something unexpected emerged in the process. Where the common narrative positions AI as a tool of extraction—taking from artists without permission—this project inverted the relationship. Nothing was being stolen. Something was being excavated. The machine became an instrument for recovering what was already mine, lost to time and circumstance, carried only in the unreliable architecture of memory.

The experience taught uncomfortable lessons about the nature of creative collaboration with technology. The machine resists the singular. It wants to give you what it has seen a thousand times before—the composite, the average, the statistically probable. To produce something genuinely personal requires iteration upon iteration, an ongoing battle to overcome the gravitational pull of training data. The prompts that finally worked were not descriptions of images but negotiations with a system that preferred to give me something else.

Where this leaves the broader questions of appropriation and originality, of style and ownership, remains uncertain. But this much became clear: for those willing to wrestle with it, the technology can serve rather than supplant. The choice is whether to accept what the machine offers by default, or to master it sufficiently to extract what was always yours to begin with.