zmime.com
Category · Image

Image prompts

Describe the shot like a photographer or art director, and the model stops guessing.

  • Photoreal
  • Illustration
  • Product
  • Style control
  • Aspect ratio
01 — Overview

Why these prompts look the way they do

Image models respond to nouns, materials and camera language far more than to praise words. "Beautiful, amazing, high quality" adds noise; "85mm, f/1.8, overcast window light, matte ceramic" adds signal. Think of a prompt as a shot list, not a wish.

Order matters too. Most image models weight the earliest tokens most heavily, so lead with the subject, then the action, then the setting, then the light, then the lens, then the style. Keep aspect ratio and version flags at the very end where the tool expects them.

02 — Anatomy

The shot-list prompt structure

  • Subject One clear subject with concrete attributes: age, material, colour, condition.
  • Action / pose What is happening. Static subjects need a pose, moving ones need a verb.
  • Environment Where it sits, plus two or three supporting details — not a whole world.
  • Light Source, direction, quality: "hard side light", "overcast", "single softbox at 45°".
  • Camera Lens length, aperture, distance, angle. This drives realism more than any style word.
  • Style / medium Photography, gouache, 3D render, risograph — plus an era or process if relevant.
  • Technical flags Aspect ratio, quality, seed, negatives. Tool-specific, always last.
03 — The prompts

5 copy-ready image prompts

Replace everything in [brackets]. Keep the constraint lines — they do more work than the descriptive ones.

Nothing matches that word. Clear the field to see every prompt.

Prompt 01

Photoreal portrait with controlled lighting

Editorial or team headshots where you need consistency across shots.

  • portrait
  • photoreal
Template · plain text 
Editorial portrait of a [age] [description] person, [expression], seated three-quarter turn toward camera, hands relaxed.

Wardrobe: [garment, fabric, colour].
Setting: [location], shallow background of [2-3 details].
Light: large softbox 45° camera-left, silver bounce fill camera-right, soft falloff on the background, catchlights visible.
Camera: 85mm lens, f/2.0, eye level, focus on near eye, natural skin texture with visible pores.
Colour: neutral white balance, muted [palette], gentle film grain.

--ar 4:5 --style raw
Prompt 02

Product shot on seamless background

Ecommerce hero images and marketplace listings.

  • product
  • ecommerce
Template · plain text 
Studio product photograph of [product], [material and finish], [colour], centred, upright, no props.

Surface: seamless [colour] sweep, subtle contact shadow directly beneath.
Light: two softboxes at 45° for even wrap, single strip light behind for edge separation, no visible reflections of the studio.
Camera: 100mm macro, f/8, dead-on eye level, whole product in frame with 12% margin on all sides.
Detail priority: crisp [logo/texture/seam], accurate colour, no dust.

Negative: extra products, text, watermarks, hands, harsh specular blowouts.
--ar 1:1
Prompt 03

Flat vector illustration for a brand system

Blog headers and feature icons that need to look like a set.

  • illustration
  • vector
Template · plain text 
Flat vector illustration of [concept, described as a concrete scene rather than an abstraction].

Style: geometric shapes, 2px consistent line weight, no gradients, no outlines on background shapes, generous negative space.
Palette: exactly these colours — [#hex], [#hex], [#hex], plus white. No other colours.
Composition: single focal object centred, two supporting elements, flat horizon, no perspective.
Mood: calm, explanatory, editorial.

Negative: 3D shading, drop shadows, photorealism, text, faces with detailed features.
--ar 16:9
Prompt 04

Scene with cinematic atmosphere

Key art, article headers and mood boards.

  • scene
  • cinematic
Template · plain text 
Wide cinematic still: [subject] in [environment], [time of day].

Action: [what is happening right now]
Atmosphere: [weather / haze / dust / rain], visible light shafts, layered depth with foreground silhouette, midground subject, distant background.
Light: [source] as key, [colour] practical lights in the midground, deep but detailed shadows.
Camera: 35mm anamorphic, f/2.8, low angle, subject occupying lower third.
Grade: teal shadows, warm highlights, filmic contrast, slight halation.

Negative: text, logos, distorted anatomy, cluttered frame.
--ar 21:9
Prompt 05

Style-consistent image set

Generating several images that clearly belong together.

  • consistency
  • series
Template · plain text 
Generate a set of [n] images that share one visual system.

FIXED ACROSS ALL IMAGES (repeat verbatim in each prompt):
Medium: [e.g. matte gouache on textured paper]
Palette: [3-4 hex codes]
Light: [single description]
Camera / framing: [single description]
Level of detail: [simple, mid, high]

VARIES PER IMAGE:
1. [subject 1 + one action]
2. [subject 2 + one action]
3. [subject 3 + one action]

Write each final prompt in full rather than referring back to the fixed block, keep word order identical between prompts, and reuse the same seed where the tool supports it.
04 — Craft notes

What separates a good prompt from a wasted run

Do this 05

  • Lead with the subject; put aspect ratio and flags last.
  • Use camera language (lens, aperture, angle) for realism.
  • Name light source, direction and quality in every prompt.
  • Lock palette with hex codes when the image must match a brand.
  • Reuse identical wording plus a fixed seed for a consistent series.

Not this 05

  • Piling on quality words like "8k, ultra detailed, masterpiece".
  • Describing three subjects and expecting all three to be accurate.
  • Relying on the model to spell text inside the image.
  • Mixing incompatible styles ("photoreal watercolour anime").
  • Changing five variables at once when a result is nearly right.
05 — Questions

Asked often, answered plainly

Why does the model ignore part of my prompt?

Long prompts dilute attention. If a detail matters, move it earlier and cut something else. Two accurate elements beat six approximate ones, and negative prompts help only for things that keep appearing uninvited.

How do I get the same character in multiple images?

Fix the character description word-for-word, reuse the seed, and use the tool’s reference features — character reference in Midjourney, or an input image where supported. See the avatar category for a full workflow.

Is negative prompting always useful?

Only for recurring problems. Adding a long default negative list to every prompt tends to flatten results. Add negatives reactively, one at a time, when you can see the artefact you want gone.

06 — Sources

Read the primary documentation

Vendors change flags, limits and defaults often. Where this site disagrees with official docs, the docs win.

07 — Tool packs

Same work, phrased for your tool

  1. 01

    Prompt for Midjourney

    Midjourney rewards concrete nouns, camera language and a short parameter tail. Adjective stacking does very little.

    • Image model
     5 prompts
  2. 02

    Prompt for Gemini

    Gemini is at its best when you hand it a lot of material and a precise job to do with it.

    • Multimodal model
     5 prompts