# Image to Prompt – Reverse Engineer Any Image

> Describe a reference image as a reusable prompt with its subject, light and style.

Canonical URL: https://allimageai.com/ai-tools/image-to-prompt/

Website: All Image AI

Language: English

You found an image you like but cannot describe it. This works backwards from it — and returns a structured prompt broken into subject, composition, lighting, lens, palette and style, phrased for the model you plan to use, rather than one vague sentence.

## Default pipeline

1. Upload a reference
2. Analyse subject / light / lens / style
3. Produce a structured prompt
4. Adapt phrasing to the target model
5. Send to the generator

## How to use

1. Upload the reference image.
2. Choose which model you want to recreate it with.
3. Send the prompt to the generator.

## Supported models

- [GPT Image](https://allimageai.com/models/gpt-image.md): OpenAI; Active.
- [Nano Banana 2](https://allimageai.com/models/nano-banana-2.md): Google; Active.

## Output

- Format: Text prompt
- Can be sent straight to the generator

## Frequently asked questions

### Can it reproduce the original exactly?

No. The prompt captures style and composition, not pixels.

### Will it tell me which model made the image?

No. Identifying the model behind an image is not reliable, so we do not guess.

## Image handling

Uploaded images are sent to the configured inference service for processing. You can delete a generation from your history.

## Related tools

- [AI Image Generator](https://allimageai.com/ai-tools/ai-image-generator.md)
- [AI Image Editor](https://allimageai.com/ai-tools/ai-image-editor.md)