Free tool ยท runs in your browser ยท nothing uploaded

AI Image Prompt Generator

Describe the picture in your head field by field and get a prompt any image model can use.

Image models reward specifics. "A cat" gets you a generic cat; "a gray tabby asleep on a sunlit windowsill, soft morning light, shallow depth of field, watercolor" gets you a picture you might actually keep. The hard part is remembering all the dimensions a good image prompt covers. This free generator lays them out as a form so you can fill in what you know and skip the rest.

Choose a subject, then a style such as photoreal, cinematic, watercolor, anime, pixel art, flat illustration, 3D render, or ink sketch. Add lighting, camera and composition, mood, a color palette, extra details, things to avoid, and an aspect ratio. The tool assembles a single prompt you can copy into any image generator, including Image Gen in ChatUp. It runs in your browser and nothing is uploaded.

How to use it

  1. Describe the subject: who or what is in the picture and what it is doing.
  2. Pick a style, then set lighting, camera or composition, mood, and a color palette.
  3. Add details worth insisting on, list anything to avoid, and choose an aspect ratio.
  4. Copy the finished prompt into your image generator and adjust after the first result.

The parts of a strong image prompt

Most image prompts that disappoint are missing a dimension rather than a detail. The subject is there, but nothing says how it is lit, where the camera sits, or what the picture should feel like, so the model picks defaults. Naming each dimension, even briefly, moves the result toward the picture you imagined.

Order matters a little. Put the subject first, the style second, and the technical notes after, since most models give earlier words more weight. Keep the whole thing readable; a wall of comma-separated keywords is harder for you to edit and not obviously better for the model.

  • Subject and action: the thing in the frame and what it is doing
  • Style: photoreal, cinematic, watercolor, anime, pixel art, flat illustration, 3D render, ink sketch, oil painting, vintage film photo
  • Lighting: golden hour, overcast, studio softbox, neon, candlelight, backlit
  • Camera and composition: close-up, wide shot, low angle, centered, rule of thirds, shallow depth of field
  • Mood: calm, tense, playful, melancholy, epic
  • Palette: muted earth tones, pastel, high-contrast black and white, teal and orange
  • Negative: what to leave out, such as text, extra hands, blur, or watermarks
  • Aspect ratio: square for avatars, 16:9 for banners, 9:16 for phone wallpapers and stories

Choosing a style

Style is the single word that changes an image most, so it is worth trying two or three. Photoreal and cinematic suit product shots and anything meant to pass as a photo. Watercolor, ink sketch, and flat illustration read as hand-made and work well for blog headers and slides. Anime and pixel art carry strong genre associations; 3D render sits between illustration and photo and is popular for icons and mascots.

If you are not sure which suits you, the art style quiz linked below narrows it down, and the ChatUp blog has a guide to AI image styles.

Frequently asked questions

Which image generators does this work with?
Any that accept a text prompt, including Image Gen inside ChatUp. The output is descriptive text; the aspect ratio is added as a short --ar note at the end, so if your generator has its own aspect ratio setting, delete that note and use the setting instead.
Is my prompt uploaded?
No. The generator runs entirely in your browser. Nothing you type is sent to a server.
What is a negative prompt?
A list of things you do not want in the picture, such as text, watermarks, or extra limbs. Some generators take it as a separate field; this tool writes it into the prompt as a plain sentence so it works everywhere.
Do I need to fill in every field?
No. Subject and style matter most. Add the rest when the first result is close but not quite right.

Keep going

More free tools

Do it in ChatUp

Now run it on a real model.

Paste what you built here into ChatUp and choose GPT, Claude, Gemini, Grok or DeepSeek for the job. One account, same history on web, iPhone and Android.