For the photorealism class, I used two of my own recent photographs — a night-lit shed and a solitary bench — as tests to see how accurately Adobe Firefly could recreate them from text prompts. Using ChatGPT to refine the descriptions, I produced AI versions that look convincingly real; in fact, I prefer the AI version of the bench for its subtle tonal atmosphere.

This experiment raises deeper questions. These images depend on photographic skill but are not photographs. I can ethically use my own work as source material, but what if someone else used my images to generate theirs? AI brings new creative possibilities — and new responsibilities — as we redefine what originality and authorship mean in a photographic world increasingly shaped by algorithms.

The RPS Digital Imaging Group recently ran a light-hearted competition on creative uses of AI, inviting members to produce photorealistic or fantasy images. I chose to extend my Tolkien-inspired mining series with two imagined scenes — The Fields of Mordor and Khazad-dûm. Using ChatGPT to craft descriptive prompts and Adobe Firefly to generate visuals, I refined each image before final editing in Photoshop.

These works aren’t photographs in the traditional sense, but they retain a photographic quality — shaped by choices of light, composition, and mood. AI, rather than replacing photography, offers photographers a new way to visualise and express imagination — expanding the boundaries of what a photographic image can be.