Short answer: after the subject, name the lens, the light and the angle, one of each, the way a photographer would brief a shoot. "A woman in a park" gives you a stock photo; "a woman in a park, 85mm lens, golden hour, shallow depth of field, eye level" gives you a portrait.
When your prompt names no camera, the model has to guess one, and the guess tends to be the safe, average picture: flat light, centred subject, everything sharp. Camera words are the quickest way to pull it somewhere else.
Which words change the picture most?
One from each row is enough. Subject first, camera words after.
| Kind | Words to try | What you get |
|---|---|---|
| Lens | 35mm | A wider view with more of the place around the subject |
| Lens | 85mm | A portrait look, the background soft and pushed back |
| Lens | macro | Very close detail: textures, drops, small things filling the frame |
| Light | golden hour | Low, warm side light and long shadows |
| Light | studio light, softbox | Even, clean light for products and headshots |
| Light | overcast | Soft light with no hard shadows |
| Angle | low angle | The subject looks taller and stronger |
| Angle | overhead, top-down | Flat-lay food and desk shots |
| Angle | close-up | Face or detail filling the frame |
What does a full prompt look like?
Plain:
Like a photographer:
Google's prompt template for photo-like images in Gemini uses the same order: the shot and the subject in a setting, then "[Description of the light]. Shot from a [camera angle] with a [lens type]." The same camera words are worth adding in Midjourney, Flux and GPT Image too. The full order (subject, setting, style, light, camera, format) is in image prompts that work in Midjourney, Flux and Nano Banana.
What goes wrong?
- Five lenses at once. "35mm, 85mm, wide angle, macro" contradicts itself; the model picks one or blends them. One lens word.
- Camera words before the subject. Lead with the person or thing the picture is about, so the camera words describe it rather than replace it.
- Light that fights the scene. "Golden hour" in "a neon-lit night street" gives a muddy mix. Pick light that could happen there.
Where can you check this?
Where do you find the right camera words?
Borrow them. Prompt Gallery shows 558 tested image prompts in 9 categories, each next to the picture it made, so you can see which lens and light words gave which look. Copy one, swap in your subject, and keep the camera words. More on Gemini in particular: Gemini image prompt tips.
