How to copy text from a photo or screenshot
How do I copy the text out of a photo or screenshot?
Save your favourite tools
Create a free iBuildPDF account to keep your favourite tools and find them quickly anytime.
Free account. The PDF tools themselves never need one.
Short answer
Upload the picture to Image to Text, leave the language on Detect automatically and press Apply: the words come back on the page, ready to copy. For a page you want to edit as paragraphs, use JPG to Word instead. Both read the picture with OCR, so check names and numbers against the original.
Why you cannot just select it
On screen, text in a picture looks exactly like text in a document, but it is stored in a completely different way. A document stores characters. A photo or screenshot stores only coloured dots that happen to form letters, and nothing in the file knows that a shape is an “a”.
The same page, with and without text
- A digital PDF. On screen, indistinguishable from the one on the right.
- Underneath it: character codes, each with a font and a position. Searchable, selectable, convertible.
- A scanned PDF. The same page, and the same appearance.
- Underneath it: a grid of light and dark samples. No letters, no words, nothing to search.
OCR (optical character recognition) is the step that looks at those shapes and decides which letters they are. The result is real text you can select, search and paste. A scanned PDF is the same problem — a picture of each page inside a PDF — and OCR PDF adds the recognised text to the file itself.
Text, Word or a searchable PDF?
- Image to Text gives you the words as plain text, shown on the page with a Copy button. Best for an address, a quote, a reference number or a paragraph from a screenshot.
- JPG to Word rebuilds the text as paragraphs in an editable document. Best for a letter or a printed page you need to change.
- OCR PDF keeps a scan looking exactly as it does and adds invisible text behind it. Best for archiving, when the picture itself matters.
None of the three recreates tables, columns or logos. They recover the words.
Choosing the language
Recognition compares shapes with what the letters of a language look like, so the language matters. Detect automatically first works out the script — Latin, Arabic, Cyrillic, Chinese and others — and then reads in the most likely language for it, using the language you browse this site in as a hint between languages that share a script. The result says which language was used.
If the text comes back as nonsense, the guess was wrong: choose the language from the list and run it again. Nothing is lost by trying twice.
What happens to the picture
Reading text in a picture needs an OCR engine and language data far too large to run well in a browser, so these tools send the image over an encrypted connection to our own server. It is read there and deleted when the job finishes, and the text is sent back only to you. If the picture shows something you would rather not upload — a card number, a password on screen — crop it out first, or type the part you need.
Tools this article covers
Sources
Primary documentation for the claims above.