AI image prompts for every model and style

Dutch-style still life of pears, a pewter jug and a half-peeled lemon on dark cloth
Vast salt flat at dawn with pastel sky reflected in thin pools and one distant figure
Orbital greenhouse with glowing green interiors and a supply ship docking over a dark planet

Prompt hub7 prompt pages

The short answer

A good AI image prompt names the subject, its setting, how it's lit, the camera and lens, the composition, the mood and the aspect ratio, in plain words. The twelve prompts on this page show that pattern, each one written for ChatGPT, Gemini and Midjourney, and the packs under this hub reuse it for one subject at a time.

AI image prompts by topic

7 pages: each one is a full pack you can copy from, not a teaser.

  • Image pack22 prompts

    Pictures of yourself

    20 looks that keep your face in a selfie, 2 about-me ideas that need no photo at all, and a…

  • Image pack20 prompts

    Best friends photo prompts

    20 ideas that keep each face from a photo of you, tips for a group of up to 4, and a flat…

  • Image pack20 prompts

    Caricature prompts

    20 ideas that keep your face, 3 dials for level, props and style, a fill-in for any job, plus…

  • Image pack21 prompts

    Realistic photo prompts

    Skin texture, light, lens and a candid frame, with 21 to copy and the edits for each of 3 tools.

  • Image pack20 prompts

    Logo design prompts

    20 fill-in AI logo prompts by style and trade, tips for text that the AI spells right, and the…

  • Image pack20 prompts

    Social media image prompts

    20 ideas for feed posts, Stories, pins and thumbnails, each at its ratio, with a spot left for…

  • Writing list40 prompts

    Picture prompts for kids

    40 scenes split by grade, K to 6, each with a question to ask, plus a safe way to get the…

Pick an aspect ratio for the prompts on this page

Choose a ratio and every starter prompt on this page takes it. The frames in this chart all share one width, so you can compare their heights at a glance.

Aspect ratio

Eight aspect ratios at the same widthOutlines of 1:1, 4:5, 3:4, 2:3, 3:2, 16:9, 9:16 and 21:9 frames, all 1,080 pixels wide and centred. The chosen ratio is filled and labelled with its pixel size.Pick a ratio to fill its frame
Aspect ratios with pixel sizes at 1,080 pixels wide
RatioShapePixels at 1,080 wideMegapixels
1:1square1,080 x 1,0801.2
4:5portrait1,080 x 1,3501.5
3:4portrait1,080 x 1,4401.6
2:3tall portrait1,080 x 1,6201.7
3:2wide1,080 x 7200.8
16:9widescreen1,080 x 6080.7
9:16vertical1,080 x 1,9202.1
21:9ultra-wide1,080 x 4630.5

No ratio chosen, so every prompt keeps the ratio it already has. Pick one to see its frame at 1,080 pixels wide.

Not written into the prompts yet. Pick a ratio and it is added to the prompts on this page.

  • The sizes are computed from the ratio at 1,080 pixels wide and rounded to the nearest pixel. Each image tool sets its own output sizes, so treat this as a shape guide.
  • Tall ratios such as 9:16 and 2:3 give the most height at a fixed width, and 21:9 gives the least.

The best AI image prompts to start with

12 to try first. Switch the model once. Every card follows.

Show every prompt for

Product shot of a hand-thrown mug

Open space on one side makes the image usable for adverts and web banners.

ChatGPT image

Create a photorealistic image: a hand-thrown speckled ceramic mug of steaming black coffee on a raw oak table. Setting: a quiet kitchen corner with a folded linen cloth. Light: soft window light from the right with a gentle highlight along the rim. Camera: macro lens-style lens moderate depth of field, slightly raised angle. Composition: mug on the lower-left third with open space on the upper right. Mood: slow morning. Aspect ratio 4:5. Extra detail: {{extra}}. No text, logos or watermark.

Gemini

Generate a photorealistic image of a hand-thrown speckled ceramic mug of steaming black coffee on a raw oak table. Setting: a quiet kitchen corner with a folded linen cloth. Light: soft window light from the right with a gentle highlight along the rim. Camera: macro lens-style lens moderate depth of field, slightly raised angle. Composition: mug on the lower-left third with open space on the upper right. Mood: slow morning. Aspect ratio 4:5. Include this extra detail: {{extra}}. Don't add text, logos or a watermark. No text, letters, numbers, logos or watermarks.

Midjourney

a hand-thrown speckled ceramic mug of steaming black coffee on a raw oak table, a quiet kitchen corner with a folded linen cloth, soft window light from the right with a gentle highlight along the rim, macro lens-style lens moderate depth of field, slightly raised angle, mug on the lower-left third with open space on the upper right, slow morning, {{extra}} --ar 4:5 --raw --no text, logo

How to use
  1. Copy it with the button above.
  2. Fill the blanks in {{double braces}} with your own details.
  3. Paste it into your image AI and generate.
  4. Then ask for one change at a time, not a new prompt.

Before you run “Product shot of a hand-thrown mug” for the first time, the lesson on how to use AI image prompts walks through copying, filling and sending it.

Close food shot of a ramen bowl

Steam and a named angle are what separate appetising food shots from flat ones.

ChatGPT image

Create a photorealistic image: a bowl of tonkotsu-style ramen with a soft egg, scallions and nori. Setting: a dark wooden counter, steam rising, chopsticks resting on a stone rest. Light: warm overhead pendant light with a faint side fill. Camera: standard lens, shallow depth of field, 45-degree angle. Composition: bowl filling two thirds of the frame, steam catching the light. Mood: comforting. Aspect ratio 4:5. Extra detail: {{extra}}. No text, logos or watermark.

Gemini

Generate a photorealistic image of a bowl of tonkotsu-style ramen with a soft egg, scallions and nori. Setting: a dark wooden counter, steam rising, chopsticks resting on a stone rest. Light: warm overhead pendant light with a faint side fill. Camera: standard lens, shallow depth of field, 45-degree angle. Composition: bowl filling two thirds of the frame, steam catching the light. Mood: comforting. Aspect ratio 4:5. Include this extra detail: {{extra}}. Don't add text, logos or a watermark. No text, letters, numbers, logos or watermarks.

Midjourney

a bowl of tonkotsu-style ramen with a soft egg, scallions and nori, a dark wooden counter, steam rising, chopsticks resting on a stone rest, warm overhead pendant light with a faint side fill, standard lens, shallow depth of field, 45-degree angle, bowl filling two thirds of the frame, steam catching the light, comforting, {{extra}} --ar 4:5 --raw --no text, letters, watermark

Rainy reading nook interior

Level verticals and a named lens prevent the stretched look common in AI interiors.

ChatGPT image

Create a photorealistic image: a reading nook in a window bay with a velvet armchair, stacked books and a floor lamp. Setting: grey rain on the window, a wool throw over the chair. Light: soft grey daylight with a warm lamp glow in the corner. Camera: ultra-wide architectural lens, level vertical lines. Composition: the armchair on the right third, window light leading the eye. Mood: cosy and still. Aspect ratio 3:2. Extra detail: {{extra}}. No text, logos or watermark.

Gemini

Generate a photorealistic image of a reading nook in a window bay with a velvet armchair, stacked books and a floor lamp. Setting: grey rain on the window, a wool throw over the chair. Light: soft grey daylight with a warm lamp glow in the corner. Camera: ultra-wide architectural lens, level vertical lines. Composition: the armchair on the right third, window light leading the eye. Mood: cosy and still. Aspect ratio 3:2. Include this extra detail: {{extra}}. Don't add text, logos or a watermark. No text, letters, numbers, logos or watermarks.

Midjourney

a reading nook in a window bay with a velvet armchair, stacked books and a floor lamp, grey rain on the window, a wool throw over the chair, soft grey daylight with a warm lamp glow in the corner, ultra-wide architectural lens, level vertical lines, the armchair on the right third, window light leading the eye, cosy and still, {{extra}} --ar 3:2 --raw --no text, letters, watermark

Vast salt flat at dawn with pastel sky reflected in thin pools and one distant figure

Salt flats at dawn

A single small figure gives scale, which is the cheapest way to make a landscape feel big.

ChatGPT image

Create a photorealistic image: a vast salt flat at dawn with thin pools of water reflecting a pastel sky and one small distant figure. Setting: empty horizon, cracked salt crust in the foreground. Light: low pink and orange light with mirror-like reflections. Camera: ultra-wide lens, deep focus, low angle just above the surface. Composition: horizon on the upper third, figure placed off-centre. Mood: vast and still. Aspect ratio 16:9. Extra detail: {{extra}}. No text, logos or watermark.

Gemini

Generate a photorealistic image of a vast salt flat at dawn with thin pools of water reflecting a pastel sky and one small distant figure. Setting: empty horizon, cracked salt crust in the foreground. Light: low pink and orange light with mirror-like reflections. Camera: ultra-wide lens, deep focus, low angle just above the surface. Composition: horizon on the upper third, figure placed off-centre. Mood: vast and still. Aspect ratio 16:9. Include this extra detail: {{extra}}. Don't add text, logos or a watermark. No text, letters, numbers, logos or watermarks.

Midjourney

a vast salt flat at dawn with thin pools of water reflecting a pastel sky and one small distant figure, empty horizon, cracked salt crust in the foreground, low pink and orange light with mirror-like reflections, ultra-wide lens, deep focus, low angle just above the surface, horizon on the upper third, figure placed off-centre, vast and still, {{extra}} --ar 16:9 --raw --no text, letters, watermark

Overhead spiral staircase

Symmetry and a named viewpoint give a poster-like image with very little prompt.

ChatGPT image

Create a photorealistic image: an overhead view looking down a white spiral staircase with a curved brass handrail. Setting: a tall stairwell with a skylight above. Light: soft light from the skylight, gentle gradients on the curves. Camera: ultra-wide lens, deep focus from the top landing, perfectly symmetrical. Composition: centred spiral, handrail forming the main curve. Mood: graphic and calm. Aspect ratio 1:1. Extra detail: {{extra}}. No text, logos or watermark.

Gemini

Generate a photorealistic image of an overhead view looking down a white spiral staircase with a curved brass handrail. Setting: a tall stairwell with a skylight above. Light: soft light from the skylight, gentle gradients on the curves. Camera: ultra-wide lens, deep focus from the top landing, perfectly symmetrical. Composition: centred spiral, handrail forming the main curve. Mood: graphic and calm. Aspect ratio 1:1. Include this extra detail: {{extra}}. Don't add text, logos or a watermark. No text, letters, numbers, logos or watermarks.

Midjourney

an overhead view looking down a white spiral staircase with a curved brass handrail, a tall stairwell with a skylight above, soft light from the skylight, gentle gradients on the curves, ultra-wide lens, deep focus from the top landing, perfectly symmetrical, centred spiral, handrail forming the main curve, graphic and calm, {{extra}} --ar 1:1 --raw --no text, letters, watermark

Dutch-style still life of pears, a pewter jug and a half-peeled lemon on dark cloth

Painterly still life with pears

Naming the old-master lighting (one window, deep shadows) gets the style without naming an artist.

ChatGPT image

Create a photorealistic image: a Dutch-style still life of pears, a pewter jug and a half-peeled lemon on dark cloth. Setting: a dark tabletop and a plain deep brown background. Light: single window light from the left with deep shadows. Camera: portrait lens, moderate depth of field. Composition: objects grouped in a loose triangle, lemon peel curling over the edge. Mood: quiet and painterly. Aspect ratio 4:5. Extra detail: {{extra}}. No text, logos or watermark.

Gemini

Generate a photorealistic image of a Dutch-style still life of pears, a pewter jug and a half-peeled lemon on dark cloth. Setting: a dark tabletop and a plain deep brown background. Light: single window light from the left with deep shadows. Camera: portrait lens, moderate depth of field. Composition: objects grouped in a loose triangle, lemon peel curling over the edge. Mood: quiet and painterly. Aspect ratio 4:5. Include this extra detail: {{extra}}. Don't add text, logos or a watermark. No text, letters, numbers, logos or watermarks.

Midjourney

a Dutch-style still life of pears, a pewter jug and a half-peeled lemon on dark cloth, a dark tabletop and a plain deep brown background, single window light from the left with deep shadows, portrait lens, moderate depth of field, objects grouped in a loose triangle, lemon peel curling over the edge, quiet and painterly, {{extra}} --ar 4:5 --stylize 200 --no text, letters, watermark

Portrait of a fictional woodworker

Hands and a task give an invented character more life than a face alone, and no real person is involved.

ChatGPT image

Create a photorealistic image: a fictional elderly woodworker with weathered hands and rolled sleeves, holding a half-finished chair leg. Setting: a cluttered workshop with hand tools on the wall. Light: north window light with fine sawdust in the air. Camera: portrait lens, shallow depth of field, eye level. Composition: head and hands in frame, the chair leg in the lower third. Mood: focused and warm. Aspect ratio 4:5. Extra detail: {{extra}}. No text, logos or watermark.

Gemini

Generate a photorealistic image of a fictional elderly woodworker with weathered hands and rolled sleeves, holding a half-finished chair leg. Setting: a cluttered workshop with hand tools on the wall. Light: north window light with fine sawdust in the air. Camera: portrait lens, shallow depth of field, eye level. Composition: head and hands in frame, the chair leg in the lower third. Mood: focused and warm. Aspect ratio 4:5. Include this extra detail: {{extra}}. Don't add text, logos or a watermark. No text, letters, numbers, logos or watermarks.

Midjourney

a fictional elderly woodworker with weathered hands and rolled sleeves, holding a half-finished chair leg, a cluttered workshop with hand tools on the wall, north window light with fine sawdust in the air, portrait lens, shallow depth of field, eye level, head and hands in frame, the chair leg in the lower third, focused and warm, {{extra}} --ar 4:5 --raw --no text, letters, watermark

Fashion editorial in a parking garage

A single bold colour against grey concrete gives an editorial feel without clutter.

ChatGPT image

Create a photorealistic image: a fictional model in a structured rust-orange coat standing very still. Setting: an empty concrete parking garage with painted lines. Light: hard overhead fluorescent light mixed with warm side bounce. Camera: standard lens, moderate depth of field, straight-on. Composition: full body centred, strong receding lines. Mood: stark and editorial. Aspect ratio 3:4. Extra detail: {{extra}}. No text, logos or watermark.

Gemini

Generate a photorealistic image of a fictional model in a structured rust-orange coat standing very still. Setting: an empty concrete parking garage with painted lines. Light: hard overhead fluorescent light mixed with warm side bounce. Camera: standard lens, moderate depth of field, straight-on. Composition: full body centred, strong receding lines. Mood: stark and editorial. Aspect ratio 3:4. Include this extra detail: {{extra}}. Don't add text, logos or a watermark. No text, letters, numbers, logos or watermarks.

Midjourney

a fictional model in a structured rust-orange coat standing very still, an empty concrete parking garage with painted lines, hard overhead fluorescent light mixed with warm side bounce, standard lens, moderate depth of field, straight-on, full body centred, strong receding lines, stark and editorial, {{extra}} --ar 3:4 --raw --no text, letters, watermark

Macro dew on a spider web

Backlight is what makes droplets sparkle, so it's named first.

ChatGPT image

Create a photorealistic image: morning dew on a spider web strung between grass stems. Setting: a meadow at sunrise with a soft green background. Light: backlit by low sun so each droplet glows. Camera: macro lens, moderate depth of field. Composition: web diagonal across the frame, droplets sharp in the centre. Mood: delicate and fresh. Aspect ratio 3:2. Extra detail: {{extra}}. No text, logos or watermark.

Gemini

Generate a photorealistic image of morning dew on a spider web strung between grass stems. Setting: a meadow at sunrise with a soft green background. Light: backlit by low sun so each droplet glows. Camera: macro lens, moderate depth of field. Composition: web diagonal across the frame, droplets sharp in the centre. Mood: delicate and fresh. Aspect ratio 3:2. Include this extra detail: {{extra}}. Don't add text, logos or a watermark. No text, letters, numbers, logos or watermarks.

Midjourney

morning dew on a spider web strung between grass stems, a meadow at sunrise with a soft green background, backlit by low sun so each droplet glows, macro lens, moderate depth of field, web diagonal across the frame, droplets sharp in the centre, delicate and fresh, {{extra}} --ar 3:2 --raw --no text, letters, watermark

Isometric illustration of a tiny bakery

Naming the flat style and the palette gives a consistent icon-like set across several prompts.

ChatGPT image

Create an image: an isometric illustration of a tiny corner bakery with a striped awning, a chimney and a delivery bicycle, flat vector style in a cream, terracotta and sage palette. Setting: a small street tile with a lamp post and a bench. Light: soft flat lighting with gentle long shadows. Camera: isometric view at 30 degrees, no perspective distortion. Composition: the building centred on a square tile with space around it. Mood: friendly and tidy. Aspect ratio 1:1. Extra detail: {{extra}}. No text, logos or watermark.

Gemini

Generate an image of an isometric illustration of a tiny corner bakery with a striped awning, a chimney and a delivery bicycle, flat vector style in a cream, terracotta and sage palette. Setting: a small street tile with a lamp post and a bench. Light: soft flat lighting with gentle long shadows. Camera: isometric view at 30 degrees, no perspective distortion. Composition: the building centred on a square tile with space around it. Mood: friendly and tidy. Aspect ratio 1:1. Include this extra detail: {{extra}}. Don't add text, logos or a watermark. No text, letters, numbers, logos or watermarks.

Midjourney

an isometric illustration of a tiny corner bakery with a striped awning, a chimney and a delivery bicycle, flat vector style in a cream, terracotta and sage palette, a small street tile with a lamp post and a bench, soft flat lighting with gentle long shadows, isometric view at 30 degrees, no perspective distortion, the building centred on a square tile with space around it, friendly and tidy, {{extra}} --ar 1:1 --stylize 250 --no text, letters, watermark

Monoline lighthouse logo mark

Asking for no text and no gradients prevents the garbled lettering and the effects that make logos unusable.

ChatGPT image

Create an image: a minimalist monoline logo mark of a lighthouse inside a circle, single stroke weight, flat navy on white. Setting: a plain white background, no shadows or gradients. Light: flat, no lighting effects. Camera: vector-style, front-on. Composition: the mark centred with generous margin. Mood: calm and trustworthy. Aspect ratio 1:1. Extra detail: {{extra}}. No text, logos or watermark.

Gemini

Generate an image of a minimalist monoline logo mark of a lighthouse inside a circle, single stroke weight, flat navy on white. Setting: a plain white background, no shadows or gradients. Light: flat, no lighting effects. Camera: vector-style, front-on. Composition: the mark centred with generous margin. Mood: calm and trustworthy. Aspect ratio 1:1. Include this extra detail: {{extra}}. Don't add text, logos or a watermark. No text, letters, numbers, logos or watermarks.

Midjourney

a minimalist monoline logo mark of a lighthouse inside a circle, single stroke weight, flat navy on white, a plain white background, no shadows or gradients, flat, no lighting effects, vector-style, front-on, the mark centred with generous margin, calm and trustworthy, {{extra}} --ar 1:1 --raw --no text, gradient, shadow

Orbital greenhouse with glowing green interiors and a supply ship docking over a dark planet

Orbital greenhouse concept art

One hard light source and a small object for scale are the two cheapest ways to make concept art read as huge.

ChatGPT image

Create an image: concept art of a rotating orbital greenhouse seen from outside, with glowing green interiors and a small supply ship docking. Setting: deep space with the dark curve of a planet below. Light: harsh sunlight on one side and deep black shadow on the other. Camera: wide-angle cinematic wide view. Composition: greenhouse ring crossing the frame, ship small at the lower right for scale. Mood: hopeful and quiet. Aspect ratio 21:9. Extra detail: {{extra}}. No text, logos or watermark.

Gemini

Generate an image of concept art of a rotating orbital greenhouse seen from outside, with glowing green interiors and a small supply ship docking. Setting: deep space with the dark curve of a planet below. Light: harsh sunlight on one side and deep black shadow on the other. Camera: wide-angle cinematic wide view. Composition: greenhouse ring crossing the frame, ship small at the lower right for scale. Mood: hopeful and quiet. Aspect ratio 21:9. Include this extra detail: {{extra}}. Don't add text, logos or a watermark. No text, letters, numbers, logos or watermarks.

Midjourney

concept art of a rotating orbital greenhouse seen from outside, with glowing green interiors and a small supply ship docking, deep space with the dark curve of a planet below, harsh sunlight on one side and deep black shadow on the other, wide-angle cinematic wide view, greenhouse ring crossing the frame, ship small at the lower right for scale, hopeful and quiet, {{extra}} --ar 21:9 --stylize 300 --no text, letters, watermark

What goes in an AI image prompt?

An AI image prompt holds 7 parts: subject, setting, light, camera and lens, composition, mood and the frame's aspect ratio, written in plain words in that order. Every pack on this hub reuses the same seven words, so once you know them you can edit any prompt on the site.

  1. Subject. One main thing, described concretely.
  2. Setting. Where it is, and what's around it.
  3. Light. Direction, softness and colour. That's what does the most work.
  4. Camera and lens. A lens length and an angle, such as 85 mm (about 3.3 inches) at f/2.
  5. Composition. Where the subject sits and what leads the eye.
  6. Mood. One or two words, such as calm or dramatic.
  7. Aspect ratio. 4:5, 3:2, 16:9 or whatever the use needs.

The AI image prompt tips guide goes deeper, and it's a quick read.

AI image prompt examples for any model

This hub has 12 AI image prompt examples, including a portrait, a product shot, a landscape and a poster-style logo mark, each written for ChatGPT, Gemini and Midjourney. Switch the model tab once and every card follows, so you don't rewrite a prompt for each tool, and you won't lose the seven parts when you do.

Each subject that needs its own rules has a pack under this hub. Self-portraits made from your own selfie have their own pack: ChatGPT prompts for pictures of yourself. Logos follow different rules (flat shapes, few colours, short text), which is why the prompt for logo design pack is separate. Posts and Stories need a set ratio and caption space, covered card by card in the AI image prompts for social media pack.

Two or more friends in one frame need a face line per person, and the prompts for friends pack writes it for you. Exaggerated portraits with props and an illustration style sit in the caricature prompt pack. Photos that should pass as real, with lens, light and skin texture spelled out, are in the AI prompt for a realistic photo pack. Teachers looking for pictures that start a story should use the picture writing prompts for kids list, which is about classroom writing, not image making.

How do ChatGPT, Gemini and Midjourney read an image prompt?

ChatGPT and Gemini read an image prompt as plain sentences and accept a reference photo, while Midjourney reads short phrases plus parameters such as --ar 4:5 and --no. The seven parts don't change; only the wording around them does.

  • ChatGPT image generation and Gemini (Nano Banana): natural sentences work best, and you can ask them to edit an uploaded photo as well as create one. Say what to keep and what to change.
  • Midjourney: a short list of comma tags, then parameters: --ar sets the ratio and --no names what to leave out. Midjourney's default model has been V8.2 since 24 July 2026, so the prompts here don't need a version flag. See the Midjourney prompts hub.
  • Keep your face: when you upload a photo of yourself, say to keep the face exactly as it is.

For Gemini photo looks built on one uploaded portrait, the Gemini AI photo editor hub groups them by occasion. Midjourney parameters such as --ar and --stylize are set for you in the Midjourney prompt builder.

Aspect ratio chart for AI images

The 6 aspect ratios AI image tools use most are 1:1, 4:5, 2:3, 3:2, 9:16 and 16:9: an Instagram feed post takes 4:5, Stories and Reels take 9:16, and YouTube and slides take 16:9. A ratio is width to height, and an image model won't guess it, so it needs the ratio in the prompt text or in a parameter.

Six aspect ratios, a common use for each and how the builder writes them
RatioSize at 1,080 px wideCommon useMidjourneyChatGPT and Gemini
1:11080 x 1080Square feed post, profile picture--ar 1:1square (1:1)
4:51080 x 1350Portrait feed post--ar 4:5portrait (4:5)
2:31080 x 1620Tall pin, portrait print--ar 2:3tall portrait (2:3)
3:21080 x 720Camera-style photo, blog image--ar 3:2landscape (3:2)
9:161080 x 1920Stories, Reels and Shorts--ar 9:16tall vertical (9:16)
16:91080 x 608Video thumbnail, YouTube, slides--ar 16:9wide landscape (16:9)

Heights are rounded to the nearest pixel. Social size guides agree on 4:5 for portrait feed posts and 9:16 for full-screen vertical video. Midjourney V8.2 only takes whole-number ratios, so you can't type 1.39, and its own docs call shapes wider than 2:1 experimental. Other models' supported sets change between releases, so check a model's documentation before you rely on an unusual shape. Square 1:1 is the safe default, and it's what Midjourney uses when you leave --ar out.

How to write an AI image prompt in 5 steps

To write an AI image prompt, name the subject first, add setting, light and lens, set the composition and mood, then end with the aspect ratio, and change one part per rerun. Here's the routine the cards follow:

  1. Pick the pack. Open the one for your subject, or start from a card on this page.
  2. Fill the 7 parts. Write one plain phrase for each; light deserves the most detail.
  3. Set the ratio. Match it to where the image will live, using the chart.
  4. Run it. Paste it into the model whose tab you picked.
  5. Adjust one part. Change a single phrase per rerun, so you can tell what moved the result.

Vague style words like "beautiful" add little; a named light source and a lens add a lot.

Which AI images shouldn't you generate?

The AI images you shouldn't generate are images of real people without their consent, children from uploaded photos, celebrities, trademarked characters and fake documents; the prompts on this site leave all five out. They use invented subjects or your own photos only. Check each tool's terms before you use images commercially, since they're not all the same and they've changed before.

Editing a photo you already have, instead of making one from text, is the job of the Nano Banana pro prompts hub. When the picture has to move, the AI video prompt hub adds camera motion, duration and sound to the same seven parts.

Prompt for AI image generator FAQ

Still stuck? Build it in Prompt Studio. It'll score each part as you go.

What is the best prompt for an AI image generator?

The best prompt for an AI image generator gives one phrase per part: subject, setting, light, lens, composition, mood and ratio. A 7-part prompt leaves the model fewer gaps to fill with guesses, and there's no single magic line that beats it.

What is a good AI prompt for a photo?

A good AI prompt for a photo names a real camera look, such as "35mm, f/2.8, soft window light", plus the subject and a 4:5 or 3:2 ratio. The realistic photo pack has 21 of them.

What are the best prompts for the ChatGPT image generator?

The best prompts for the ChatGPT image generator are full sentences that say what to keep and what to change, with the aspect ratio written at the end. Each card on this page has a ChatGPT version in its model tab, so you don't have to rewrite one.

Are AI image prompts copyrighted?

No, a short AI image prompt is usually not protected on its own, but rules for AI images differ by country, so check the law where you publish before you sell an image.

Make your own in Prompt Studio

Answer a few questions. The prompt's scored as it's built. It's free, and nothing leaves your browser.