Recognition

Turn screenshots into Markdown

Drop in a screenshot and get structured Markdown back — headings, lists, tables and fenced code blocks, not a flat wall of text.

Tables preserved//

Code blocks//

Any text language

Source image

Drop an image here, paste from your clipboard, or choose a file.

PNG · JPG · WEBP · up to 3 MB

Markdown

What it handles well

Built for screenshots of documents, tables, terminal output and code. Handwriting, heavily skewed photos and dense multi-column layouts are where it starts to guess — check those results carefully.

System Capabilities

Detailed Specs

Structure, not just text

Headings, lists, blockquotes and tables come back as real Markdown syntax instead of loose lines.

Tables stay tables

Tabular screenshots are rebuilt as Markdown tables you can paste straight into a README.

Code keeps its fences

Code screenshots return as fenced blocks with a language hint, ready to paste into docs.

Failed reads cost nothing

If the image has no readable text, your daily quota is returned instead of spent.

Operational Manual

Workflow

Add your image

Drop a file, paste a screenshot from your clipboard, or pick one from disk.

Extract

A vision model reads the layout and rewrites it as Markdown.

Check and copy

Edit anything that looks off, then copy the Markdown or download it as an .md file.

Start Converting Now

Troubleshooting

Support

Which image formats work?

PNG, JPG and WEBP, up to 3 MB per image. Larger files do not improve accuracy — they only slow the request down.

Do I need an account?

Yes. Every conversion runs a vision model with a real compute cost, so it is tied to an account with a daily free allowance.

What happens if the recognition fails?

If we cannot find readable text, the conversion does not count against your daily quota. You will see why it failed instead of a blank result.

Does it handle handwriting?

Sometimes, but that is where accuracy drops most. Printed screenshots of documents, tables and code are the reliable case.

Are my images stored?

No. The image is sent for recognition and discarded — nothing is written to disk.

Does it translate the text?

No. Text comes back in the original language of the image.

Next step