Turn an Image into a Prompt
Get an English description you can edit, copy or use to create a new image.
Describe image
Example

Description
A bicycle beside a flower shop on a wet cobblestone street, red flowers in terracotta pots, stone facades and soft natural light.
Image to Prompt: A Starting Point for Your Next Visual
The image to prompt tool turns visible content into an editable English description. Use this image to prompt generator to put references into words in Flux Context.
Build a Visual Brief with Image to Prompt
- Subject and surroundings
- Name the main object or scene and where it sits. Add important objects the description missed and remove irrelevant detail.
- Composition and light
- Describe framing, viewpoint, lighting and colors when they matter. A caption may not capture every spatial relationship.
- The change you want
- Turn the description into your own brief: change the weather, materials or setting rather than copying the text unchanged.
From Reference to Editable Text with Image to Prompt
- 1
Upload a Clear Reference
Choose one photo with a readable subject and composition. You do not need to write a prompt first.
- 2
Generate the Description
Review quoted credits and process the input photo. The result is text, not a new picture.
- 3
Edit, Copy or Create
Edit text in the result box, copy it, or send it to generation. Creating a new visual is a separate action.
Choose What You Want to Describe
One clear composition
Use a single photo rather than a collage when you need one coherent brief. Tiny lettering may be omitted.
A visual reference, not hidden settings
The visual reference does not reveal its seed, model settings or original prompt.
Make the Text Your Own
- Correct omitted objects, invented details or misleading spatial descriptions.
- Keep the visual properties you want and explicitly name the ones you want to change.
- Descriptions are English; edit or translate the text for your intended use.
Questions About This Tool
Does Image to Prompt Recover the Original Generation Prompt?
No. It produces a description of visible content. The original words, seed and generation settings cannot be recovered from this description.
What Language Does Image to Prompt Output?
The current workflow generates English descriptions, including when the interface is Spanish. You can edit the text before using it.
Will Image to Prompt Recreate an Identical Visual?
No. A caption does not encode every visual detail, and different models interpret text differently.
Does Use in Creation Generate Immediately in Image to Prompt?
No. It places the edited text in the workspace draft. Review the prompt and price before starting a task.
Which Files Can I Upload?
You can upload PNG, JPG, or WebP files up to 4MB and 40 megapixels. Transparent PNGs are fully supported, and files are processed at their uploaded resolution without pre-compression downscaling.
When Are Credits Deducted, and What Happens If Processing Fails?
Credits are only deducted when you click the action button to process the image. Adjusting sliders, uploading, or previewing settings never costs credits. If a task encounters a server error or timeout, your credits are refunded immediately and automatically.
Can I Use Tool Outputs in Other Editors or for Video Generation?
Yes. Any processed cutout, upscaled image, or restored photo can be opened directly in Image Editing, added as a reference in Multi-Reference generation, or sent to our AI Video studio as a starting frame.

