Image Prompt Generator

Describe the picture, pick the model, get the prompt in the form that model reads.

More AI tools

Prompt library

How to use the image prompt generator

Start from a few words or from a picture. The generator fills in what an image model needs and leaves out what it ignores.

  1. 101Pick the image modelChoose Midjourney, Nano Banana, GPT Image, FLUX, Seedream and Stable Diffusion. Each one gets its own syntax: parameters for Midjourney, sentences for Nano Banana, tags and a negative prompt for Stable Diffusion.
  2. 202Describe or uploadType the subject, place and mood, or upload a picture to write the prompt from it. Under Options set the aspect ratio (1:1, 4:5, 3:2, 16:9, 9:16) and a style.
  3. 303Refine, copy or renderMake it shorter or more detailed, try it for another model, then copy it or run it in the studio.

What is the image prompt generator?

It turns a rough description into a prompt an image model follows: subject, setting, composition, light, lens and style, ordered and phrased for the model you pick.

  1. 016 image models, one box

    Midjourney: comma phrases, then --ar and style parameters. Nano Banana: full sentences: subject, scene, light, camera. GPT Image: plain description, exact text in quotes, layout. FLUX: subject first, rich natural language, no negatives. Seedream: concise scene, style and text in quotes. Stable Diffusion: weighted tags plus a negative prompt.

  2. 02Start from a picture

    Upload, paste or drop an image and the generator writes the prompt that recreates it. Add a note such as make it night to change one thing.

  3. 03Aspect ratio and style built in

    Pick the aspect ratio and a style (auto, photo, cinematic, illustration, 3d render, anime and product shot). Midjourney gets them as --ar and parameters, the other models in words.

  4. 04Render it in the studio

    Send the prompt to the studio and run it on Nano Banana, GPT Image, FLUX or Seedream. New accounts get 60 free credits.

Image prompt generator FAQ

Which models it writes for, how the syntax differs, and how to turn a picture into a prompt.

1Which image models does it write prompts for?
Midjourney, Nano Banana, GPT Image, FLUX, Seedream and Stable Diffusion. Pick one in the menu under the text box; the prompt is rewritten in that model's syntax.
2How is a Midjourney prompt different from a Nano Banana prompt?
Midjourney reads comma-separated visual phrases followed by parameters such as --ar 16:9 and --stylize. Nano Banana and GPT Image read full sentences, and any text that should appear in the picture goes in double quotes. Stable Diffusion uses weighted tags like (tag:1.2) and a separate negative prompt.
3Can I get a prompt from an image I already have?
Yes. Click Upload in the text box, or paste or drop a picture, and the generator writes a prompt that recreates it for the model you picked. For style-only or subject-only prompts use Image to Prompt.
4Does it add words like 8k or masterpiece?
No. Filler tags rarely change the picture on current models. The prompt describes what you actually want to see: subject, place, light, lens and style.
5Can I change the prompt after it is written?
Use the buttons under the result: Shorter, More detail, Add an example or More formal rewrite the prompt you have, and Try for rewrites it for another tool. Each refinement counts as one prompt.
6Is it free?
Yes. You get 10 free prompts a day on each device without an account, and 20 a day with a free account. Paid plans remove the daily limit.
7Do you keep what I type or upload?
Ideas and prompts written on this page are not saved to an account. They are sent to our AI model provider to write the prompt and are not shown to anyone else. Save keeps a copy in your own browser only.