AI image describer: what's in the picture, in words
Upload an image and get three things back: a clear description, alt text you can paste, and a prompt that would generate something like it, all from one upload.
Drop an image here
or paste it, or pick a file
No image loaded.
One difference worth knowing.Most of the image tools here never send your file anywhere. This one has to, because reading an image takes a vision model that can't run in a browser. Your image goes to a processing provider for that one job and is not stored by us afterwards: no copy of the image and no copy of the description. If you would rather nothing left your device, the PDF converter and the other browser tools are unaffected.
Image to prompt, alt text, or just an explanation
Image to prompt
Found a picture you want more of? The prompt it writes names the subject, setting, style, lighting, lens and palette, in the order image models read them. Paste it into any generator, or send it straight to ours.
Alt text
A single sentence under 125 characters that says what the image shows, without the "image of" filler. A solid first draft for a blog, a store listing or a social post.
Understanding an image
A plain description of what is there, including any text in the picture, quoted exactly. Useful for a screenshot in a language you're working through, or a chart you need to summarise.
The prompt is written the way image models are documented to read one. Google's prompt and image attribute guide starts from subject, context and style, and the rest of the prompt adds lighting, lens and palette on top. If you would rather build a prompt from an idea than from an image, the AI image prompt generator does that.
The alt text follows the rules in the W3C's alt decision tree: describe what the image is for, keep it short, and skip "image of". Screen readers already say that part.
Frequently asked questions
How do I turn an image into a prompt?
Upload the image and press Describe. Along with the description you get a prompt written for text-to-image models: subject first, then setting, style, lighting, lens and colour. Copy it, or use Generate with this prompt to open it in Leaxor's image generator.
Will the prompt recreate the exact same image?
No, and no describer can promise that. A prompt describes an image; it doesn't encode one. You will get something with the same subject, mood and style, and each model interprets it a little differently. That is usually what people want, since an exact copy of someone else's image is rarely the goal.
Is the alt text good enough to publish?
It is a strong first draft. Good alt text depends on why the image is on the page, which only you know. The W3C's alt decision tree is the best short guide to that judgement: a decorative image needs empty alt, and a chart needs its point, not its colours.
Is it free?
Yes, within a daily allowance, and it needs a free account. Guest sessions can't run it, because every run is tied to a real account.
Is my image uploaded?
Yes. Describing an image needs a vision model that can't run in a browser, so the image goes to a processing provider for that one job. It is not stored by us afterwards, and neither is the description.
Will it identify the people in a photo?
No. It describes what is visible — clothing, expression, pose, setting — but it is instructed not to guess who someone is from their face.
Other image tools
Got the prompt. Now make the picture.
Leaxor puts ten image models behind one prompt box, from Flux 2 Pro to Imagen 4 Ultra. Try the same prompt on a different model instead of arguing with one.
See the image generator