Developing your photography

Photography has always involved a process of development, an interpretive stage that sits somewhere between capture and presentation. From the chemical alchemy of early darkrooms to the algorithmic polish of modern software, photographers have consistently shaped their images in pursuit of clarity, mood, and impact. My recent heavily cropped, high-ISO image of a Dartford Warbler was, by any technical standard, far from ideal: noisy, soft, and lacking the crisp detail one hopes for in wildlife photography. So I did what photographers have always done, I tried to improve it. Only this time, the tool I used was generative AI.

And that’s where things become less straightforward.

Using a generative system to develop the image did not simply refine what was already there. It introduced detail, feather structure, tonal nuance, and even subtle changes in the bird’s posture, that were never actually captured by my camera. The result is, undeniably, a more visually compelling image. But it is also, just as undeniably, no longer a faithful representation of the moment I witnessed. It looks better, but it looks less real. And perhaps more importantly, it is no longer entirely mine.

This raises a question that feels both new and strangely familiar: where does one draw the line?

Photographers have been manipulating images since the medium’s inception in the 19th century. Early practitioners retouched negatives, combined exposures, painted directly onto prints, and later added colour by hand. In that sense, the idea of altering an image beyond what was strictly captured is not new. But those interventions, however creative, were constrained, by the artist’s own hand, by time, by visible technique. They were interpretive, not generative.

What distinguishes generative AI is not just what it can do, but how it does it. It does not enhance captured data so much as replace or supplement it with statistically plausible invention. When it improves a bird’s plumage, it is not recovering lost detail from noise; it is synthesizing what such detail ought to look like, based on patterns learned from vast datasets. The image shifts from being an index of a real event, light hitting a sensor at a specific time and place, to something more ambiguous: a hybrid of observation and prediction.

At that point, the nature of authorship becomes blurred. The photograph begins as mine, my timing, my framing, my decision to press the shutter. But the final image is co-authored by a system trained on countless other images, many made by people I will never know. The more the AI contributes, the less the image can be said to originate solely from my act of seeing. It becomes less a photograph and more an interpretation, or perhaps an illustration, of one.

And yet, the impulse behind using such tools is entirely consistent with photographic history: the desire to realise the image as we felt it, rather than as the camera imperfectly recorded it. That tension, between documentation and expression, has always been present. What generative AI does is amplify it, and in doing so, forces us to confront it more explicitly.

There are, perhaps, a few ways to navigate this. One is to maintain a distinction between images that remain grounded in captured data and those that incorporate generative elements, not necessarily as a moral boundary, but as a matter of clarity. Another is to be transparent about the process, acknowledging when an image has moved beyond traditional development into synthesis. And a third is to accept that intent matters: an image meant as a document carries different expectations than one intended as an aesthetic object.

There is also an irony that is difficult to ignore. Like many others, I have spent decades putting creative work into the world, photographs, writing, videos, and music, all of which now form part of the vast, largely invisible substrate on which these systems are trained. In that sense, the AI’s output may contain diffuse echoes of my own past work, just as it contains those of countless others. But that does not restore authorship; it only underscores how collective and impersonal the process has become.

Rest assured, this was a one-off experiment. None of the photos I share will be genAI, they will all have been my shutter, and my shutter alone.