Making These Photographs

As a child I used to hold my mum’s slides up to the window; the tiny positive frames made me feel like I was in another world. I’ve never let go of that feeling; it has shaped my research ever since.

I bought a job lot of slides on eBay. One of them caught my attention: a Kodachrome of an unnamed woman sitting on a couch in window light.

I lost it almost immediately.

I never found out who she was, but she stayed with me. And because I didn’t know her story, I made one up.

The photographs on this site illustrate a novel, not a real family. The project began with a real image and developed into an archive of things that never happened.

I photographed Lila’s chapters on a research trip to New York, and my family photographed the New Zealand chapters. Others began as text. I stripped scenes down to keywords and generated them with Google’s Gemini image model running inside Adobe Firefly. Using the initial digital file as a starting point, I then composited, retouched, and graded the work.

Every image on this site has its own row in a separate key: what it is, how it was made, and where it came from.

Making Photographs for a Fictional Family

As a photographer, I couldn’t help visualising the photographs the characters took as I wrote about them. But I cannot photograph Francis’s New York in 1948, Darla’s Rochester, or Sophie’s nineties. If those chapters were going to have photographs, I would have to make them.

Generation was the start. I worked on each image by hand, printed it on paper that matched the character’s era, and rephotographed it on a copy stand.

Fake Materiality

Rephotographing the images mattered because I wanted them to become physical objects. To do this, I matched the paper, lighting and copy stand so the set would hold together. I felt that printing and rephotographing an image would make a more convincing fake than the digital file alone. Generated images fail: their lighting doesn’t make sense, and their surfaces are too smooth. The grain and texture of my paper masked these flaws, making them look convincing enough.

The 1948 images don’t document anything real; nothing stood in front of the lens. By rephotographing them on a copy stand, I made a real print of a fictional event, but it didn’t make the event real.

The Slide — Remade

The slide at the centre of the novel is slightly different. The original was lost. When the book needed a picture of it, it was still lost, so I made one.

I generated the image, printed it digitally onto positive film, mounted it in a 1973 Kodachrome mount, had it inscribed by hand, and rephotographed it on the copy stand.

The inscription reads: Rose, Locust St, 6/28/73. Rose is a character. The street is in the novel, and I chose the date. With the Kodachrome slide that started all of this gone, the slide in the archive is a prop, a stand-in for the missing original.

Sophie’s Archive

To build Sophie’s archive, I used my own photographs from living in Hell’s Kitchen and the Lower East Side during the 1990s to shape her fictional world. Where a chapter required a specific photograph, I used a personal photograph if it fit. If no photograph supported the prose, I generated digital fragments to match her story.

Sophie’s kitchen, shown above, is the clearest example. I built the generated frame from two photographs stuck in an old album of my apartment on Rivington Street: the same room, the same light, most of the same objects. Then I altered it to fit the story rather than my memory. The microwave on my fridge became a boombox, because the manuscript needed somewhere for Sophie to play a cassette. I even got to do the dishes thirty years later.

The question was which version belonged to the archive. One of my photographs might reflect my memory accurately but not serve Sophie’s story. A generated image might be historically fake but serve her story better.

Real Friction

Lila’s photographs follow her path through the novel. They are tied to an Instagram account I keep under her name: @lilarae___. The New Zealand chapters show a harbour and a neighbourhood my family knows first-hand.

Placing real photographs into an archive creates a strange friction. They show real places, yet the story reframes their history. Can a photograph be factually real without validating the fiction built around it?

This is what interests me about archiving. An image doesn’t arrive in a box in the loft with its meaning already attached. Someone decided which ones to keep and which to throw away. It might have names and places etched into the mount or written on the reverse, but the human connection, the reason for keeping it, is lost. The family album is a series of choices made by the person who put the photographs there.

In the novel, the characters inherit photographs, objects, oral histories, and absences. This mirrors how most of us inherit family history: as physical fragments we hold tightly while imagining moments we never experienced. Lila holds a photograph of a stranger named Rose. Sophie swaddles her baby in a blanket inherited from someone she doesn’t remember.

My approach is more deliberate: I made these characters’ inherited photographs by hand, from nothing. I am interested in what a photograph must do, materially, to feel like it belongs to someone, even when the viewer knows it was made to do just that.

Objects

The objects in this project came the same way the lost slide did: an old wardrobe key, a Kodak Retina camera, a hand-knit baby blanket from the 1970s. I bought each on eBay from a stranger’s clearance. Collecting became a quiet, ongoing practice, shaped by what each chapter required, and folded into the fictional archive.

What I Don’t Know

I no longer have access to the full Firefly prompt history. What survives are fragmented filenames produced by the software: the first few words of each prompt, cut off partway through. Because of this, what you see here is an account of my practice, not a complete record.

There are deeper questions I can’t resolve. Every generated image here came from Google’s Gemini model. Adobe’s assurances about how its Firefly models were trained don’t apply here, and Google doesn’t fully explain what its image model learned from. I can’t verify any of it. Keeping photographers’ names out of my prompts doesn’t sidestep that. This is the stage of the project where I have the least control and the least authority to make claims. I’d rather say that than pretend I can account for an algorithm I can’t see.

The Line I Didn’t Cross

Some chapters carry no photograph at all. They’re set in conflicts far from anything I’ve lived through, among unnamed people whose suffering has nothing to do with mine. Inventing the visual memory of fictional characters is one thing. Turning real historical trauma into a generated image and asking a reader to look at it is another. I didn’t want to make that claim. Those chapters stay blank on purpose. The absence is part of the archive, not a gap in it.

What This Is

The only real thing is the practice: a copy stand, two lights, a grey card read at the start of every session, the camera locked at the same height, matched paper stock, boxes of objects, photographs made and remade until they held up. The labour is real, as are the objects; the family is not.

I was building the conditions for a fictional past to feel inherited, for a reader who knows, the whole time, exactly what they’re looking at.

What Remains

The photograph that started all this is still gone. I never found out who she was, who took the picture, who kept it, or how it ended up in a job lot on eBay. All I could do was build a world around her: a family, a history, photographs, objects, rooms and absences.

The real question was never whether these photographs are real. It’s what you do with a person you can’t recover.