Note: All the images featured in this article are generated by Chat GPT based on original photographs, which I have taken.

Ai generated image based on my photograph: a man floating in the ocean with flippers on; a yellow pontoon in the background.

I’ve been playing with AI text-to-image generation a lot recently; whether it’s seeing how AI tools can translate my photographs into illustrative or painterly renditions, or using prompts to create imagery for visuals I use at my gaming table.

In all of this, AI image generation has taught me a valuable lesson: sometimes, it pays to let go.

The uncanny, evocative, and often unpredictable results of AI image generation suggest something fundamental about the creative process: that spontaneity, chance, and the unknown are not just inevitable, but essential ingredients in compelling visual storytelling.

Ai generated image based on my photograph: Diptych -- a misty morning at a wetlands; a shearer checking wool.

As photographers, we spend a lot of time trying to get things just right. The right light, the right lens, the perfect framing, the perfect moment. There’s a certain satisfaction in control — in knowing our tools and bending the world (or at least our little piece of it) into a beautiful frame.

But here’s the thing: we also know that some of our best shots weren’t planned at all.

They happened because the light hit differently than expected. Because someone walked into frame at the perfect moment. Because we missed the shot we thought we wanted and found something better on the way.

Ai generated image based on my photograph: a family running into the sea; a man with a wave ski at the beach.

Photographers, especially in this digital era, are told to control variables — light, composition, exposure, timing. The pursuit of “perfection” is often the pursuit of predictability. Yet, history shows us that some of the most powerful photographs were born not from planning, but from a moment’s chaos, a split-second decision, or an unforeseen gesture. From Henri Cartier-Bresson’s “decisive moment” to the street photography of Vivian Maier and Robert Doisneau, chance has always had a place at the creative table.

Text-to-image AI — tools like DALL·E, Midjourney, and others seem, at first glabce, like the antithesis of traditional photography. You type in a prompt, and et voilà, an image appears! No camera, no waiting for the light, no shutter click. These tools operate under a curious paradox: the more specific the input, the more predictable the output — but often, it’s the vague or poetic prompts that produce the most stirring images. What I’ve discovered is that relinquishing control over every detail invites the model to surprise us with unexpected juxtapositions, novel aesthetics, or uncanny atmospheres.

Let’s take a prompt like: “The last memory of summer, fading like a dream.”

It doesn’t say anything about what should be in the picture — no mention of people, places, or objects. But that’s exactly why it works. Feed that into an image generator, and you might get a warm, hazy field bathed in late sunlight… a child’s toy left in the grass… maybe a leaf caught mid-fall, golden and soft.

“The last memory of Summer, fading like a dream.” – image generated with ChatGPT.

It’s not just a picture — it’s a feeling. And here’s where things start clicking for photographers. That feeling is something we chase all the time, even when we don’t have the words for it. We wait for it in the frame. We sense it when a photo has that unspoken something. That moment where the technical and the accidental collide.

Using AI like this is weirdly freeing. You can control it to a point, sure — but push too hard, and it becomes literal, stiff, uninspired. The best results often happen when you don’t micromanage every detail. You leave room for interpretation. For accidents. For a little bit of poetic chaos.

أفكار وتجارب.

What AI Image Generation has Taught me as a Photographer