Guide · Image

Get the prompt of an image, then remake it with AI

A photo spotted on Pinterest, a poster, a screenshot: you want the same look without guessing the words. The Beemm Vision Chrome extension turns the image into a prompt, then sends it to Generate. Here is the full method, from the analysis to your own version, in four steps.

5 min readBeginnerUpdated 9 November 2026

In short

To get the prompt of an image: with the Beemm Vision Chrome extension, hover the image and click Prompt (or right-click, then Beemm Vision), run the analysis, then click To Beemm Vision. The prompt, the image and its format land in Generate, where you pick the model that remakes it.

An image from the web
Beemm Vision extension
Analysis
Gemini 2.5 Flash by default
Languages
EN, FR
Send
Generate
Prompt, then a new image
How it works: an image from the web, the extension’s analysis, a new image in Generate.Open the studio

What the analysis does

A vision model looks at the image and describes it for a generator: subject, style, light, framing. The extension returns this prompt in English and in French, with shot details and a hex palette. The prompt is not the author’s: it is a faithful description, written to get a close result. On the canvas, the Image Describer node does similar work on an image already in Beemm Vision.

EN · FR

the prompt comes in both languages

20

latest analyses kept on your computer

1 click

to send the prompt and the image to Generate

Examples made on Beemm Vision

Starting photo: a speckled ceramic cup on a linen tablecloth next to a croissant, in morning sun
The starting image, the one whose prompt you want: subject, low morning light, a cream and terracotta palette.
The same scene regenerated from a written description alone, with no image attached
Remade from a description alone (subject, framing, light, palette), with no image attached: the scene and the light come back, the details change.
Variant: same framing and light, a sage-green teapot and a slice of lemon cake
With the image kept as a reference and a rewritten subject: same framing, same light, a visual of your own.

The four steps

  1. Step 1

    Install the extension and sign in

    Add the Beemm Vision extension from the Chrome Web Store, then sign in with Google, with the same account as the app. The account email must be verified. Installing is free; each analysis uses credits.

  2. Step 2

    Point at the image

    Hover an image on Pinterest, Behance, Unsplash or Dribbble: the Prompt button appears. Elsewhere, right-click the image and choose Beemm Vision, or open the panel and pick "Sélectionner une image" (select an image). For part of the screen, "Capturer une zone" (area capture) draws a free selection; on a Mac, ⌘⇧C starts it while the panel is open.

  3. Step 3

    Pick the analysis model, then run

    The preview shows the image, the analysis model and its cost before anything is sent. Gemini 2.5 Flash is the default; other Gemini, GPT and Qwen models are on offer, some with an effort level. To steer the analysis, "Créer un prompt personnalisé" (custom prompt) opens instructions of up to 1,200 characters, which you can save under a name and reuse.

  4. Step 4

    Review, then remake the image in Generate

    The result opens on the prompt, in EN or FR, with the Détails (capture, light, texture, composition…) and Palette tabs. Copy copies it for another tool; To Beemm Vision opens Generate with the prompt, the image as a reference and its format. Keep the image to stay close to the original, or remove it and rewrite the subject for a version of your own. Pick the model, then run: the cost shows first.

The extension at a glance

StepWhat the extension offersTip
StartPrompt button on hover, right-click, select an image, area captureArea capture also works for a background image or a paused video
Analysis modelGemini 2.5 Flash by default; Gemini, GPT, QwenSome offer an effort level; the cost shows first
Instructions1,200 characters, 20 saved promptsTo stress the light, the material or the framing
ResultPrompt in EN or FR, Details, PalettePalette: each tone in hex, with its share and its role
SendCopy, or To Beemm VisionGenerate gets the prompt, the image as a reference and its format

Use cases

Start from a moodboard reference

An image spotted on Pinterest becomes a prompt you rework: another subject, the same light.

Reuse a palette

The Palette tab gives the tones in hex: paste them into the prompt to keep the colour mood.

Describe your own images

On the canvas, an Import node then an Image Describer node: the description comes out as text, ready to connect to an image model’s prompt.

Vary without starting over

Keep the image as a reference in Generate and change one thing in the prompt: the object, the season, the material. The method is the one in the AI photo editing guide.

Frequently asked questions

How do I get a prompt from an image?
With the Beemm Vision Chrome extension: hover the image and click Prompt, or right-click it and choose Beemm Vision. A vision model describes it as a prompt, to copy or to send to Generate to remake it.
Is the prompt I get the one that created the image?
No. The analysis does not read the original prompt: it writes a new one that describes the subject, style, light and framing to get a close result.
What about an image already in Beemm Vision?
On the canvas, connect the image (an Import node or a model output) to an Image Describer node: it returns a one-paragraph description, to connect to an image model’s prompt. Its default model is Gemini 2.5 Flash.
Why does the image land in Generate too?
To Beemm Vision sends the prompt, the image and its format. Attached as a reference, the image switches a model such as Nano Banana Pro or Seedream 5 Pro to editing: the result stays close to the original. Remove it to generate from the prompt alone.
How much does an analysis cost?
Installing is free. Each analysis uses credits from your Beemm Vision account; the cost depends on the analysis model and, for some, on the effort level, and shows before you run.
Can I remake another creator’s image?
The prompt gives you a starting point. For a visual you will publish, make it your own: change the subject, the composition or the palette. The original image remains its author’s work.

Try it on your next reference

Install the extension from the Image to prompt page, sign in with your app account, then hover an image: the prompt lands in Generate in one click.

Keep reading