Does the word "photorealistic" do anything at all?

Some. OpenAI recommends asking for it explicitly when that is the goal. But it is a weak signal compared to naming materials, texture and light, and it will lose an argument with any part of the prompt that implies a polished image.

Do camera and lens specifications actually change the render?

They change the appearance, because the model has learned what images tagged with those specifications look like. They do not simulate optics. Treat "50mm" as a look you are naming, not a lens you are mounting.

Which model should I run this on?

Either. The official example was published on both. Sunburst holds fine skin texture slightly better, which is the whole subject of this prompt, so it is the one worth the extra wait here.

Why 2:3 rather than square?

A standing figure in a portrait frame leaves room for the boat and the dog without cropping the subject. A square frame forces the model to choose between the person and the context.

GPT Image 2.5 prompt — a photorealistic portrait with the retouching turned off

Most "photorealistic" prompts fail because they ask for beauty. This one asks for pores.

An elderly sailor with weathered skin adjusts a net on a small fishing boat, his dog on the deck behind him

Settings the Get a portrait that does not look retouched prompt ran at
Mode	Model	Quality	Size	Ratio	Credits
Text to Image	Think	Standard	1K	2:3	6 cr

Published by OpenAI as an example for this model — not generated on this site. Model: gpt-image-2.5-sunburst. Source: OpenAI The published example ran at 1024x1536, a legal size we do not currently offer; the frame here is the same shape at our nearest size.

An elderly sailor on a fishing boat, shot as if on 35mm film. What makes it read as a photograph is not the word "photorealistic" — it is the four lines that follow it, naming wrinkles, pores, sun texture and faded tattoos, and the two at the end that forbid glamorisation. A model asked only for realism will give you a beautiful person under studio light, because that is what most photographs of people in its training data look like.

In full

The prompt, nothing cut

Create a photorealistic candid photograph of an elderly sailor standing on a small fishing boat.
He has weathered skin with visible wrinkles, pores, and sun texture, and a few faded traditional sailor tattoos on his arms.
He is calmly adjusting a net while his dog sits nearby on the deck. Shot like a 35mm film photograph, medium close-up at eye level, using a 50mm lens.
Soft coastal daylight, shallow depth of field, subtle film grain, natural color balance.
The image should feel honest and unposed, with real skin texture, worn materials, and everyday detail. No glamorization, no heavy retouching.

603 characters

Run it here

This exact prompt, at these exact settings

The mode and all five settings are locked to this template, so there is nothing to line up by hand. Press Run this prompt above and the panel opens filled in — prompt, mode, quality, size and frame — and your first image needs no account. If your plan sits below what this ran at, the line under the panel says so.

This example was published at Think · Standard · 1K. A free run is Fast · Standard · 1K, so expect less fine detail and softer small text. The composition is the same; the finish is not.

3 levers

What is safe to change

Most libraries publish this list and stop, which is why so many copied prompts come back worse than the original: the swappable nouns are the safe half. The three sections under it are the other half — the clauses that are doing the work, the ones that break it if you touch them, and why it is written in this order.

The texture list
Wrinkles, pores, sun texture, faded tattoos. Swap these for whatever your subject has actually lived through — calloused hands, chipped nail polish, a shirt washed too many times. The list is the realism; the word "photorealistic" is not.
The camera line
"35mm film photograph, medium close-up at eye level, using a 50mm lens" is a description of an appearance, not a physics simulation. Change the focal length and you change how much background you get and how the face flattens — but treat it as a style cue, because that is how the model reads it.
The two prohibitions
"No glamorization, no heavy retouching" is the line doing the most work per character. Delete it and the sailor gets younger, cleaner and better lit on almost every rerun.

Three ways to break it

Each one is a real failure with a reason attached. A rule without a reason is not usable.

01Adding "beautiful", "stunning" or "award-winning"
These words point at a different corpus. Award-winning portraiture is lit, posed and graded, which is the opposite of the candid register the rest of the prompt is asking for — and when a prompt contradicts itself the model does not pick a side, it averages.
02Dropping to Draft quality for the face
Skin texture is the first thing a lower quality tier spends less on. The prompt asks for pores by name, so at Draft you get a prompt whose central claim is not rendered — the picture will look smooth and the words will look ignored.
03Naming a real photographer
It replaces a specific description you control with a style average you do not, and the style average usually includes exactly the polish this prompt is trying to avoid.

Why it is written this way

Inside the Get a portrait that does not look retouched prompt

This is the first template in the library and it is here to make a point about the word "photorealistic", which is the single most over-used and least effective token in image prompting. On its own it does very little. What it competes against is everything else in the prompt, and if the rest of the prompt describes an attractive person in nice light, "photorealistic" will simply render that attractively and in nice light.

What actually produces a photograph is specificity about imperfection. Read the second line again: weathered skin, visible wrinkles, pores, sun texture, faded traditional sailor tattoos. None of those are aesthetic requests. They are a list of things a camera would have recorded and a retoucher would have removed. Naming them is how you tell the model which side of that process you want to be on.

The third line is doing something subtler. "He is calmly adjusting a net while his dog sits nearby" gives the subject a task. A person doing something is not looking at the camera, and a subject not looking at the camera is most of what "candid" means in practice. Compare the alternative — "an elderly sailor looking at the camera" — which produces a posed portrait no matter how much film grain you ask for afterwards.

The camera specification is worth understanding rather than copying. OpenAI's own guidance is explicit that camera specifications are cues for appearance and not a guarantee of physical simulation, and that is the right way to hold it: "50mm lens" is shorthand for a familiar look, not an instruction to the renderer. It is useful because it is a compressed description that the model has seen attached to millions of images, which is a more efficient way to spend tokens than describing the same compression yourself.

Finally, the two prohibitions at the end. Prompts are read as a whole, and a positive description of realism is weaker than a negative constraint on the thing that breaks it — because there is exactly one way to be unretouched and a thousand ways to be beautiful. If you take one line from this template into your own work, take that one.

4 questions

Get a portrait that does not look retouched — common questions

Does the word "photorealistic" do anything at all?: Some. OpenAI recommends asking for it explicitly when that is the goal. But it is a weak signal compared to naming materials, texture and light, and it will lose an argument with any part of the prompt that implies a polished image.
Do camera and lens specifications actually change the render?: They change the appearance, because the model has learned what images tagged with those specifications look like. They do not simulate optics. Treat "50mm" as a look you are naming, not a lens you are mounting.
Which model should I run this on?: Either. The official example was published on both. Sunburst holds fine skin texture slightly better, which is the whole subject of this prompt, so it is the one worth the extra wait here.
Why 2:3 rather than square?: A standing figure in a portrait frame leaves room for the boat and the dog without cropping the subject. A square frame forces the model to choose between the person and the context.