Six worked photo prompt examples, and why each line is there

How to read these examples

Each prompt below is written as short sentences, one decision per sentence, in a fixed order: subject, setting, moment, light, composition, lens, colour. The breakdown after each one says what that sentence is doing. Copy them, then change a single sentence at a time so you can see what each part controls. That is how a prompt becomes reusable rather than lucky.

Example one: an architectural interior

A concrete stairwell with a curved handrail. An empty modernist library, late morning. Diffused daylight from a tall window on the left, soft, with gentle shadows on the steps. Wide shot from the base of the stairs, looking up, the curve leading out of frame at the top. Wide focal length, deep focus. Concrete grey and pale oak, nothing else.

What the interior prompt is doing

The subject is a material and a shape, not an adjective. Empty is doing real work, because interiors default to people. Late morning and a tall window together fix the light without naming a mood. Looking up from the base is a geometry instruction the model can execute exactly. Deep focus keeps the architecture legible, which is the point of the image. The two-colour palette stops it drifting toward a stock rendering.

Example two: a city moment

A woman in a yellow raincoat crossing an empty street. A narrow commercial street with shuttered shops, autumn, just after rain. Overcast light, soft and even, with wet asphalt reflecting the shopfronts. Medium wide shot at eye level, the subject small and off to the right, the street receding to the left. Long focal length, compressed perspective. Grey and wet black, one note of yellow.

What the city prompt is doing

Just after rain is the most efficient sentence in the prompt: it produces reflections, a cool cast and empty pavement in three words. The subject is deliberately small, because street photographs where the person fills the frame become portraits. Compressed perspective from a long lens stacks the shopfronts, which is what makes an image read as street rather than staged. One colour note against a muted palette is a classic structure and it survives most edits.

Example three: an everyday object

A single ripe fig, split open, on a chipped ceramic plate. A plain wooden table against a dark wall. Late afternoon. A single hard light source from behind and to the right, throwing a long shadow across the table and catching the wet interior of the fruit. Close-up, slightly above eye level, the plate off-centre. Long focal length, shallow depth of field with the background falling away. Deep red, cream, dark brown.

What the object prompt is doing

Split open is the entire reason the image is interesting, so it sits in the subject sentence. Backlight from one hard source is what makes wet surfaces read as wet, and it is the choice most people miss when they light an object flatly from the front. The chipped plate is a single imperfection, and imperfections are what stop still-life prompts from looking like catalogue photography. Slightly above eye level is how food is actually looked at.

Example four: a landscape

A line of bare poplars along a flooded field. Flat farmland with a distant treeline, winter, early morning fog. Low soft light from behind the fog, no visible sun, the trees fading in tone toward the background. Wide shot at eye level, the trees running from left to right across the lower third, sky occupying the rest. Long focal length, compressed layers. Grey, brown, the palest blue.

What the landscape prompt is doing

Fog is doing the work an adjective like atmospheric would fail at, because it produces a real physical effect: aerial perspective, where distant things lose contrast. Naming that the trees fade in tone tells the model to render it rather than hoping. No visible sun prevents a generated sunburst, which is the most common intrusion in landscape prompts. Putting the subject in the lower third and giving the sky the rest is a composition decision, stated as geometry.

Example five: a plate of food

A bowl of dark lentil stew with a torn piece of flatbread resting on the rim. A worn kitchen table with a linen cloth pushed to one side. Midday, indoors, away from the window. Soft light from a large window behind and to the left, no direct sun, a faint highlight along the rim of the bowl. Close-up at a low angle, the bowl filling most of the frame, the background out of focus. Wide aperture. Deep brown, oatmeal linen, one green herb.

What the food prompt is doing

The torn bread and the pushed-aside cloth are traces of a person, which is what separates food photography from product photography. Away from the window and soft light from behind and to the left is a real setup, and it produces the gentle rim highlight that makes food look edible rather than plastic. A low angle makes the bowl feel substantial. The green herb is the single colour accent doing the same job as the yellow raincoat earlier.

Example six: a workspace

A drafting table covered in pencil sketches, a metal ruler holding one sheet flat. A small studio with a bare plaster wall, evening. A single warm desk lamp low on the left, everything beyond its reach falling into shadow. Overhead shot looking straight down, the table filling the frame, the edges dark. Wide focal length, deep focus so the sketches stay readable. Warm white paper, graphite grey, brass.

What the workspace prompt is doing

One lamp and everything else in shadow is a decision that most prompts avoid, and it is why this image would have a shape. The metal ruler is a specific object rather than clutter, and specific objects anchor a scene. Straight down is unambiguous, unlike from above. Deep focus is a deliberate contradiction of the usual instinct to blur, chosen because unreadable sketches would defeat the subject.

The pattern underneath all six

Every one of these follows the same order and the same discipline: a physical subject, a setting that supplies the light, a moment that fixes the conditions, one clearly stated light source, a geometric framing instruction, a lens behaviour chosen for a reason, and a palette of two or three colours. None of them contains a word about mood. The mood is a consequence of the other choices.

Adapting one without breaking it

Swap the subject and setting first and leave the light, composition and lens sentences alone. Those are the parts doing the photographic work, and they are the ones people delete when a prompt feels long. Once the new subject is working, change one of them at a time. If you change three sentences and the result improves, you have learned nothing you can use again.

Building your own collection

Keep the prompts that nearly worked, with a line on what actually came back. Six examples become thirty within a month if you record the near-misses, and a library of near-misses is what you adapt when a new brief arrives. A collection of only successes teaches you much less, because success rarely explains itself.

Saying what these images are

Images generated from prompts like these are not photographs, however photographic they read. The IPTC Digital Source Type vocabulary provides values that distinguish a digital capture sampled from real life from media created using generative AI, and the C2PA specification defines how that provenance travels with a file. Label anything that might be taken for a record of something real. LarpGPT provides prompts and reference material, and what you generate is yours to use and yours to disclose.

Further reading

• IPTC Digital Source Type NewsCodes
• C2PA technical specification