Image to text converter

Extract text from image files, photos, and screenshots

This converter uses optical character recognition (OCR) to find the letters in a picture and turn them into text you can edit. It works on screenshots, scanned pages, photos of documents, and pictures of signs and labels. The text appears in the box beside your image, where you can correct it before you copy it or save it.

The reading is done by Tesseract, an open source OCR engine, running inside your browser. Your image stays on your device the whole time. There is no signup and no daily limit, because no server is doing the work.

Screenshot to text: paste with Ctrl+V or Cmd+V

The fastest way to read a screenshot is to paste it, without saving it as a file first.

• On Windows, press Windows key + Shift + S, select the area you want, then press Ctrl+V on this page.
• On a Mac, press Cmd + Ctrl + Shift + 4, select the area, then press Cmd+V on this page.

You can also right-click an image on a web page, choose Copy Image, and paste it here the same way.

Copy text from image in one click

When the reading finishes, press Copy text to put everything on your clipboard, or Download .txt to save it as a plain text file. If a word came out wrong, fix it in the box first, because both buttons use whatever the box contains.

To get the text into Word or Google Docs, copy it here and paste it into your document. The converter gives you plain text, so the headings, bold type and table borders of the original do not come across.

JPG to text, PNG to text, and other formats

The converter reads JPG and JPEG, PNG, WebP, BMP, and GIF images. For an animated GIF, it reads the first frame. It does not read PDF files or iPhone HEIC photos. To read a HEIC photo, export it as a JPG, or take a screenshot of it and paste that.

Images under 1,000 pixels wide are enlarged to twice their size and read with a setting that copes better with uneven lighting. In our tests on a blurry, low-resolution purchase order, that step removed about three quarters of the character errors.

Photo to text and picture to text on your phone

On a phone, tap the box at the top of the page to pick a photo from your library or take a new one. The quality of the photo matters most. Hold the phone square to the page, fill the frame with the text, and make sure the text is sharp and evenly lit.

Because the reading happens on the phone itself, a very large photo takes longer, and an older phone with little memory can struggle. If that happens, crop the photo to the part you need and try again.

Free OCR online that runs in your browser

Many free OCR sites send your image to their server and read it there. This page reads it in your browser. The first time you use the tool, your browser downloads the OCR engine and the English and French language data, about 5 MB, from a public CDN (cdn.jsdelivr.net). After that, your browser reuses its saved copy, and your image is never uploaded.

The page counts how often the tool is used, along with the file type and size. It never records the image or the text.

Image to text in English and French

The default setting reads English and French together, which suits bilingual documents. In our tests, English on its own got only about half of the French accents right, while English + French got every one. The English and French settings read one language each and download less data the first time.

If your image is in English and full of numbers, choose English. The combined setting sometimes writes a decimal point as a comma, the way French does, so 696.53 can come back as 696,53.

Convertir une image en texte en français

Déposez une image ou collez une capture d'écran avec Ctrl+V ou Cmd+V. L'outil lit le français et l'anglais par défaut, accents compris, et le texte s'affiche, prêt à être copié. L'image est lue dans votre navigateur et n'est jamais téléversée. La première fois, votre navigateur télécharge le moteur de reconnaissance et les données de langue, environ 5 Mo, à partir d'un CDN public (cdn.jsdelivr.net).

What OCR reads well, and what it gets wrong

OCR is very good at printed text. Screenshots, typed documents and clear photos of printed pages usually come back with few or no errors. In our tests, clean printed text in English and French came back with every character right, accents included.

It is much weaker in these cases:

• Handwriting. Neat block capitals sometimes work. Joined-up cursive usually does not, so expect to retype most of it.
• Low resolution and blur. Text that is tiny, blurred or in shadow loses characters, and a badly blurred image can come back as mostly nonsense.
• Tables. A table comes back as lines of text, one row at a time, and the column layout is lost. Small marks such as decimal points are easy to misread, so 64.10 can come back as 64:10.
• Several columns. A page with two or more columns of text can come back with lines from each column mixed together. Crop to one column at a time.
• Angled photos. Text photographed at a steep angle, or on a curved page, reads worse than text shot straight on.

Check numbers carefully before you rely on them. For the best results, use a sharp, well-lit image where the text fills most of the frame.

When you need this for invoices or purchase orders every day

A free OCR tool suits the occasional screenshot or page of notes. It falls short when a team retypes purchase orders and invoices into a business system every day, because plain text still has to be split into fields and checked against your own records before anyone can use it. Our guide to text extraction from images covers that production work, from free OCR up to systems that fill in the fields and send only the uncertain ones to a person.

This tool is made by ThriveAI, which builds private AI systems for manufacturers and distributors.