Camera and lens terms for AI prompts: focal length, aperture, shots and film looks

Image models learned from enormous numbers of captioned photographs, and many of those captions came with photographic language attached. That is why a phrase like “85mm portrait, shallow depth of field” changes an image so reliably: the model has seen what those words tend to look like. This guide covers the camera and lens vocabulary that carries real visual meaning, what each term does, and which ones are mostly noise.

Focal length: how the frame feels

Focal length is the single most useful camera term, because it changes perspective, background and the sense of distance all at once.

24mm and wider

Wide-angle. Shows a lot of the scene, exaggerates depth, and stretches anything near the edges. Close to a face, it distorts features, which is occasionally useful for energy or comedy and rarely flattering. Good for interiors, landscapes, architecture and environmental portraits where the setting matters.

24mm wide-angle photo of a cramped workshop, tools covering every wall, a carpenter at the bench

35mm

The classic documentary and street-photography focal length. Close to how a scene feels when you stand in it. Includes context without obvious distortion. A safe choice when you want an image to feel candid and present.

50mm

Often described as “normal”. Neutral perspective, neither wide nor compressed. Useful when you want a plain, honest look without the model reading too much into the lens.

85mm

The traditional portrait lens. Flattering facial proportions, a background that falls softly out of focus, and a feeling of intimacy without being in someone’s face. Probably the single most effective lens term for portraits.

85mm portrait of an elderly fisherman, weathered face, harbour softly blurred behind him

135mm and longer

Telephoto. Compresses distance, so background elements look larger and closer to the subject. Isolates the subject strongly. Good for candid moments, sports, wildlife, and for making a city skyline loom behind a figure.

Macro

Not a focal length, but a lens type the models understand well. “Macro photograph” brings extreme close detail, very shallow focus and texture: dew on a leaf, the weave of fabric, an insect’s eye.

Aperture and depth of field

Aperture controls how much of the image is in focus. In prompts, the f-number is shorthand for a familiar look:

• f/1.4 to f/2: very shallow depth of field. Subject sharp, background melted into soft blur and round highlights (bokeh).
• f/2.8 to f/4: subject clearly separated, background still recognisable.
• f/8 to f/16: front-to-back sharpness. Landscapes, architecture, product shots where everything must read.

Plain language works as well as numbers, often better: “shallow depth of field”, “soft bokeh background”, “everything in sharp focus”, “deep focus”. Combining a focal length with a depth-of-field phrase is more reliable than either alone, because they reinforce the same idea.

Avoid conflicts. “f/1.4” with “everything in sharp focus” pulls in two directions, and so does asking for a soft background while putting blurry in a negative prompt; see negative prompts explained for that trap.

Shot types: how much of the subject is in frame

Film and television terms map cleanly to framing:

• Extreme close-up: a detail, such as eyes, hands or a texture.
• Close-up: head and shoulders.
• Medium close-up: chest up.
• Medium shot: waist up.
• Cowboy shot: mid-thigh up.
• Full shot / full-body shot: the whole figure, head to toe.
• Wide shot: the figure small within the setting.
• Establishing shot: the location first, with people as part of it.

Shot types are among the terms models follow most consistently. If a portrait keeps cropping off the feet, “full-body shot” fixes it faster than any amount of description.

Camera angles: where the camera is

Camera position changes composition and emotion more than most adjectives:

• Eye level: neutral, relatable.
• Low angle: the subject looks powerful, tall, imposing.
• Worm’s-eye view: extreme low angle, from the ground looking up.
• High angle: the subject looks smaller or more vulnerable.
• Bird’s-eye view / top-down / overhead: flat-lay products, food, maps, patterns.
• Dutch angle: tilted horizon, tension or unease.
• Over-the-shoulder: puts the viewer in a scene, behind one person facing another.
• Aerial / drone shot: landscapes and cities from high above.

low-angle full-body shot of a dancer mid-leap against a pale sky, 35mm

Film looks: grain, colour and processing

Film vocabulary describes the character of an image rather than its geometry:

• 35mm film photograph / analogue photo: visible grain, softer highlights, slightly imperfect colour.
• Medium format: smoother tones, very fine detail, often a squarer or 4:5-feeling frame.
• Black-and-white film, high grain: gritty documentary or street look.
• Colour negative film, warm skin tones: soft contrast and natural, slightly warm colour, common in portrait and wedding work.
• Slide film, saturated colours: punchy blues and greens, deep shadows, popular for landscapes.
• Tungsten-balanced night film with halation: red-orange glow around bright lights, a familiar night-street look.
• Expired film, light leaks: faded colours and orange streaks for a nostalgic, lo-fi feel.
• Cross-processed: shifted, unnatural colour casts.

Many models also recognise the names of specific commercial film stocks, but descriptive phrases like the ones above are more portable between models and make your intent explicit. The tools that turn images into prompts, such as our image describer, tend to describe film looks in this generic way for the same reason.

Which terms models actually respond to

Some photographic words consistently change the image; others mostly add noise.

Strong, reliable signals

• Focal length when paired with the subject (“85mm portrait”, “24mm interior”).
• Depth-of-field phrases (“shallow depth of field”, “deep focus”).
• Shot types and camera angles.
• Film and processing looks (“35mm film grain”, “black-and-white”).
• Light described photographically (“backlit”, “rim light”, “golden hour”, “overcast soft light”).

Weak or unpredictable

• Exact numbers stacked together: “ISO 100, 1/250s, f/5.6” rarely does more than “sharp daylight photo”. Shutter speed only matters when it implies a visible effect, such as motion blur or frozen action, so describe the effect instead.
• Camera body names. They carry some general “professional photo” association but very little specific to the camera.
• Generic quality words such as “8k”, “ultra HD” or “award-winning”. They describe a reputation, not a look.

The shared vocabulary works across Midjourney, Stable Diffusion, Flux and Gemini, but the formatting differs. Midjourney and Stable Diffusion are comfortable with compact phrases; Flux and Gemini prefer the same terms inside full sentences. Our Flux vs Midjourney comparison shows one scene written both ways.

Before and after

Before: a woman reading in a café, beautiful, realistic, high quality, 8k

After:  medium close-up of a woman reading by a café window, 85mm, f/1.8,
        shallow depth of field, soft overcast window light from the left,
        warm colour negative film look, fine grain

Before: a city at night, cinematic

After:  wide establishing shot of a rain-soaked city intersection at night,
        24mm from a high angle, deep focus, neon reflections on wet asphalt,
        tungsten-balanced film look with halation around the streetlights

Before: breakfast on a table, food photography

After:  top-down overhead shot of a breakfast spread on a linen tablecloth,
        50mm, f/8, everything in sharp focus, soft morning light,
        clean editorial food photography

In each case, the extra words answer questions the model would otherwise guess at: how far away is the camera, where is it, how much is sharp, and what does the image look like as a photograph. The Midjourney prompt generator and Stable Diffusion prompt generator both add this kind of camera detail when they describe an uploaded photo.

Quick reference

1. Pick a focal length for the feel: 24mm for space, 35mm for candour, 50mm for neutral, 85mm for portraits, 135mm for compression.
2. Set depth of field in words or with an f-number, and don’t contradict it elsewhere.
3. Name the shot type to control how much of the subject is in frame.
4. Say where the camera is: eye level, low, high, overhead, over-the-shoulder.
5. Add one film look if you want texture or a period feel.
6. Skip stacked settings and quality buzzwords; describe visible effects instead.

For the other half of a strong prompt, light, colour and style, see seven habits that fix generic AI images.