Free AI image prompt generator

Build strong prompts for Midjourney, Flux 2, GPT Image, Nano Banana, and Ideogram. 28 visual styles, model-aware syntax, no signup, no limits.

Step 2 of 3

Describe your image

The subject is the only required field. Every chip is optional.

Quick start (optional)
0 / 600
Advanced· all optional
Free. No signup required

How AI image prompting works

Writing an AI image prompt is not the same as writing a text prompt. Text models want a role, a task, and context. Image models want a visual brief: a stack of descriptors that tells the model what to put in the frame, what it should look like, and how it should feel.

Strong image prompts stack four components. The subject is the thing in the frame. The style and mood sets whether it's a photograph, a flat illustration, a cel- shaded anime frame, a painterly concept piece, or something else. The lighting and camera controls the visual language: golden hour rim light, an 85mm portrait lens, a low-angle cinematic wide shot. And the composition and palette decides the frame shape, colour story, and where the eye goes first.

Our generator walks you through all four, then emits the prompt in the dialect your chosen model understands: comma-tag syntax with Midjourney flags, flowing natural paragraphs for GPT Image and Flux, or Ideogram's quoted text markers for typography work.

Supported image models

Each model has its own prompt dialect. The generator emits the right syntax automatically.

Flux 2 Klein

natural language

Black Forest Labs

Free, fast, open-weights. Generates in under a second. Strong photoreal output and typography. Future versions of Promptrace will render this inline; for now, the prompt opens in BFL Playground.

Midjourney v7

tags + flags

Midjourney

Highest aesthetic ceiling for stylised illustration and art-directed photography. No public API: copy the prompt into Midjourney's web or Discord client.

Nano Banana Pro

natural language

Google

Google's current SOTA image model (Gemini 3 Pro Image). Best prompt-following, accurate text in images, multi-turn editing. Natural language prompts.

GPT Image 1.5

natural language

OpenAI

OpenAI's current flagship image model. Strong instruction following, excellent in-image text, transparent backgrounds. Pure natural-language prompts.

Ideogram 3

natural language

Ideogram

The text-in-image specialist. Best-in-class typography for posters, packaging, and logos. Natural language with explicit text emphasis.

Flux 2 Pro

natural language

Black Forest Labs

BFL's highest-quality model. 4MP output, strong photorealism and typography, supports up to 10 reference images for character consistency.

Example image prompts

Nine fully-formed prompts across the main style groups. Copy any of them as a starting point, or use the generator above to build your own.

Cinematic portrait, photoreal

GPT Image / Flux 2 / Nano Banana
A cinematic, photoreal portrait of a weathered fisherman in his sixties, standing on the deck of a small wooden trawler at dawn. Shot on an 85mm lens at f/1.8, shallow depth of field, with soft golden-hour rim light catching the salt in his beard. Muted teal and amber palette, overcast sky behind, a thin morning mist softening the horizon. Square 1:1 composition, subject centered, gaze slightly off-camera. Editorial magazine aesthetic, no text, no watermark.

Moody cinematic landscape

Midjourney v7 / Flux 2
Sweeping landscape of a lone red farmhouse on a ridge of rolling Icelandic hills, storm clouds gathering overhead, rain falling in the distance catching the last sliver of evening sun. Cinematic wide shot, anamorphic lens flare, muted palette of deep green, slate grey, and burnt sienna. Composition uses strong leading lines and rule-of-thirds placement. --ar 21:9 --s 250 --v 7

Flat vector illustration

Nano Banana / GPT Image
Flat vector illustration of a remote worker at a tidy desk by a large window overlooking a pine forest. Minimal geometric shapes, clean line weights, warm neutral palette of cream, terracotta, sage green, and navy. Subtle paper texture, soft shadows, no gradients. Editorial tech-blog style, 3:2 landscape, friendly and calm mood, no text in the image.

Isometric room, detailed illustration

GPT Image / Nano Banana
Detailed isometric illustration of a cosy writer's attic study, viewed from a 30-degree top-down angle. Wooden floor, sloped ceiling with a skylight, bookshelves overflowing, a vintage typewriter on a small desk, a cat curled on a tartan rug. Warm interior lighting spilling onto the floor, cool blue evening light through the skylight. Soft shadows, painterly textures, rich but muted palette. 1:1 square composition.

Anime-style character art

Midjourney v7 / Nano Banana
Anime-style illustration of a young cartographer in a long teal coat standing on a cliff edge, wind lifting the pages of her open atlas. Cel-shaded, bold line art, vivid highlights on her hair, soft diffuse rim light from a setting sun. Background is a mountainous coastline rendered in watercolour gradients. Studio-Ghibli influence, 4:5 portrait, no text. --ar 4:5 --s 350 --v 7

Minimalist product shot

Flux 2 Pro / GPT Image
Minimalist product photograph of a matte ceramic coffee mug on a pale linen cloth, beside a single sprig of olive branch. Soft north-facing window light, gentle shadows falling to the right, warm neutral palette of bone white, oat, and sage. 4:5 portrait framing, shallow depth of field, subject occupies lower-left third. Editorial catalogue aesthetic, no props, no text, no branding.

Logo concept, clean geometric

Ideogram 3
Clean, geometric logo mark for a fintech startup called Northkeep. A minimal shield formed from two interlocking north-arrow shapes, rendered in a single solid accent colour on a flat off-white background. Balanced negative space, subtle optical adjustments, vector-perfect curves. text: "NORTHKEEP" rendered beneath in a modern geometric sans-serif. 1:1 square, plenty of padding, no gradients, no 3D shading.

Fantasy concept art

Midjourney v7 / Flux 2 Pro
Painterly concept art of a half-buried cathedral in a vast desert of red dunes at blue hour, cracked marble columns half-swallowed by sand, a lone robed figure approaching the entrance with a lantern. Epic scale, atmospheric haze, warm firelight glow contrasting cool twilight sky, palette of ochre, burnt orange, and deep indigo. Cinematic wide shot with dramatic leading lines. --ar 16:9 --s 500 --v 7

Editorial food photography

Flux 2 Pro / GPT Image
Overhead editorial food photograph of a rustic wooden board with sliced sourdough, a small bowl of olive oil, a ramekin of flaky salt, and scattered rosemary sprigs. Natural side light from a window at 10 o'clock, soft shadows, deep warm palette of honey, terracotta, and deep green. Linen napkin partially in frame. 4:5 portrait composition, shallow depth of field on the salt, slight film grain. No text.

Frequently asked questions

What makes a good AI image prompt?

A good image prompt names four things: the subject (what's in the frame), the style (photoreal, illustration, anime, cinematic, etc.), the lighting and camera (golden hour, 85mm portrait, wide-angle), and the mood or composition. Vague prompts like 'a cat' produce generic results; strong prompts stack those four components so the model has enough signal to reproduce a specific image in your head.

Which image models does this generator support?

Seven: Midjourney v7, Flux 2 Pro and Flux 2 Klein (both from Black Forest Labs), Nano Banana Pro (Google Gemini 3 Pro Image), GPT Image 1.5 (OpenAI), Ideogram 3 (best for in-image text), and a 'Generic' universal mode. Each model uses a slightly different prompt dialect (Midjourney takes comma-tag syntax plus --flags, GPT Image and Nano Banana prefer natural paragraphs), and the generator automatically emits the right one for your chosen model.

How do Midjourney prompts differ from GPT Image or Flux prompts?

Midjourney was trained on comma-separated descriptors and responds best to tag-style prompts plus its --ar / --s / --c / --v 7 flags. GPT Image, Nano Banana, and Flux 2 are trained on natural language and respond best to flowing descriptive sentences. Ideogram sits in between but needs quoted text markers like text: "Grand Opening" to accurately render typography. Our generator picks the right dialect based on the model you select.

Is the image prompt generator free?

Yes. Building image prompts, picking styles, and deep-linking to Midjourney, Nano Banana, GPT Image, Ideogram, or BFL Playground are all free, with no signup and no daily limit. Prompt composition runs entirely in your browser; we never store your prompts on our servers.

Can I use these image prompts commercially?

The prompts produced by this tool are yours to use however you like, including commercially. Whether the resulting IMAGE is licensed for commercial use depends on the image model you generate it with, not on Promptrace. Check the commercial terms of Midjourney, Flux, GPT Image, Nano Banana, or Ideogram directly.

Do I need to know prompt engineering to use this?

No. The generator guides you through picking a subject, style, mood, lighting, camera, composition, and palette via visual chips. No memorising keywords, no staring at a blank text box. If you're new, start with a 'Quick start' preset (Photo, Illustration, Logo, or Concept art) and tweak from there.

Need a text prompt instead?

The homepage wizard builds prompts for writing, analysis, advice, and code, scored across seven dimensions.

Open the text prompt builder →