Reference·May 17, 2026·7 min

AI Image Prompt Vocabulary: A Cheat Sheet for Camera, Lighting, and Composition

The exact words that actually change what the model renders — bookmark and reuse.

Why vocabulary is the highest-leverage thing to learn

You can't prompt what you can't name. "Nice lighting" is invisible to the model — it has no concept of "nice." But "golden hour from camera-right, soft fill bounce, rim light catching the hair" gives it three unambiguous instructions it can execute.

This is a working cheat sheet of the vocabulary that actually moves the output. Keep it open while you prompt.

Camera and lens

Term	What it does
35mm film / Kodak Portra 400	Warm color cast, fine grain, photographic feel
Medium format / Hasselblad	High detail, shallow depth, editorial polish
f/1.4 / f/2.8	Shallow depth of field — subject sharp, background soft
f/8 / f/11	Deep focus — everything sharp, documentary feel
85mm portrait lens	Flattering compression, blurred background
24mm wide angle	Expansive scene, slight edge distortion
Macro lens	Extreme close-up, fine surface detail
Tilt-shift	Selective focus band, miniature effect

Lighting

Term	What it does
Golden hour	Warm, low-angle sunlight, long shadows
Blue hour	Cool dusk, deep saturation
Overcast soft light	Even diffuse light, no harsh shadows
Hard noon sun	Sharp shadows, high contrast
Rembrandt lighting	Triangle of light on cheek, dramatic portrait
Rim light / backlight	Halo separating subject from background
Soft fill from camera-left	Reduces shadow density on left side of subject
Practical lights only	Lit by visible lamps, no studio rig — cinematic realism
Bounce flash	Soft, directional, editorial portrait feel

Composition

Term	What it does
Rule of thirds	Subject placed at intersection, breathing room
Centered symmetrical	Bold, formal, Wes Anderson feel
Dutch angle	Tilted frame, tension and unease
Low angle / worm's eye	Subject feels imposing, heroic
High angle / bird's eye	Subject feels small, vulnerable, or schematic
Negative space	Large empty area, minimalist editorial
Full bleed	Image fills frame to all edges, no margin
Leading lines	Visible lines pulling eye toward subject
Foreground / midground / background layers	Depth and scale

Aspect ratio

Ratio	Use case
1:1	Instagram, album art, profile
3:2	DSLR photography, prints
16:9	Cinematic, web hero, thumbnail
9:16	Vertical mobile, Stories, Reels
2:3 portrait	Book covers, magazine, editorial poster
4:5	Instagram feed (max tall)

Color and palette

Replace "colorful" with specific palette callouts:

Restrained palette: ivory, deep teal, near-black, one orange accent — coherent editorial
Monochromatic warm earth tones — desaturated, premium
High-key — bright, low-contrast, airy
Low-key — dark, dramatic, moody
Complementary blue and orange — high contrast, cinematic
Pastel washed-out — gentle, lifestyle
Risograph two-color (red + blue) — print-shop feel, registration offset

Medium and finish

Naming the medium pins the rendering style harder than any adjective:

35mm film, slight grain, warm cast
Shot on Hasselblad, medium format detail, shallow depth
Risograph print, two-color, faint registration offset
Oil on canvas, visible brushwork, palette knife texture
Watercolor on cold-press paper, wet edges, paper bleed
3D render, Cinema 4D, Octane, soft global illumination
Pixel art, 32x32 grid, limited palette, dithering
Pen and ink, crosshatching, high contrast
Cyanotype print, deep blue, paper texture
Polaroid, faded edges, slight overexposure

Type and text rendering

When the image contains text, treat it as typographic instruction:

Headline reads "YOUR COPY" in [font style], [weight], [color], [placement]
Font styles to name explicitly: condensed grotesk, geometric sans, slab serif, brush script, monospace, transitional serif, display serif, neo-grotesque
Weight: light, regular, medium, bold, black, 100–900
Placement: top-aligned, baseline-centered, bottom-right corner, [N]px from top edge
Treatment: outlined, drop shadow, no shadow, raised letterpress, embossed, debossed

Style references that work

Skip "in the style of [famous artist]." Use medium + technique combinations:

Editorial magazine photography — clean, restrained, premium
Documentary photojournalism — natural, unposed, on-location
Brutalist graphic design — flat color, heavy type, off-grid
Swiss International Style — grid-aligned, sans-serif, restrained
Bauhaus poster — primary colors, geometric, modernist
Wabi-sabi — imperfect, asymmetric, weathered, Japanese
Memphis design — bold geometric, primary colors, 1980s
Y2K aesthetic — chrome, gradients, glossy plastic, early-2000s tech

Putting it together

Pick one term from camera, one from lighting, one from composition, one from palette, and one from medium. Five specific words usually beat fifty vague ones.

Weak: *Beautiful portrait with nice lighting and good colors.*

Strong: *85mm portrait lens at f/2.8, Rembrandt lighting from camera-right, rule of thirds, monochromatic warm earth tones, 35mm film with Kodak Portra 400 grain.*

The second prompt is shorter and produces a far more consistent image — because every word does work.

Use it without the lookup

Depikt bakes this vocabulary into every prompt it writes — type a rough idea and you get the structured output with camera, lighting, composition, and medium already specified. Or browse the library for 500 prompts that demonstrate the patterns in context.

Generate yours

Generate polished prompts in seconds.

Paste a rough idea. Get back a structured prompt that ships.

Aspect ratios for AI images: which to use for what (GPT Image 2 guide)

The single biggest reason AI images come out wrong: no aspect ratio in the prompt. GPT Image 2 defaults to 1:1 and crops everything else awkwardly. Here's the right ratio for every use case, with the exact phrase to drop into your prompt.

How-to

How to prompt GPT Image 2 for logos and brand marks

GPT Image 2 can spit out genuinely usable logo directions if you brief it like a designer. Here's the four-block structure: concept, mark type, type system, finish — plus copy-paste templates.