Guide
How PDF Text Extraction Differs From OCR
Last updated October 5, 2026
"Get the text out of a PDF" sounds like one job, but it is two different problems depending on how the PDF was made. Mixing them up is the most common reason a text tool seems to fail.
Text-layer extraction: reading text that is already there
A PDF made from a word processor, a web page or most design software contains real text: the same characters you can select and copy in a viewer. Text-layer extraction, which is what PDF to Text does, reads those characters out of the file. It is fast and exact, because nothing is being guessed.
OCR: recognising text inside a picture
A scanned document is different. Each page is a picture of a page, with no text in the file, so as far as the PDF is concerned it is the same as a photo. To get text from it you need optical character recognition (OCR), which looks at the pixels and tries to recognise letters. Running a scanned PDF through PDF to Text does not produce an error, but it returns little or no text, because there is no text layer to read. PDF to Text does not do OCR.
Where this site offers OCR
Two editors include OCR, and in both it runs locally in your browser:
- The PDF Text Editor can run OCR on a scanned page so you can find and change the recognised lines.
- The Image Text Editor uses OCR to find the words in a JPG, PNG or WebP.
The file and the recognised text are not sent to our server, and the recognition engine and its language data are served from this website rather than from a third party. Our tests found no upload of the file or the text while an editor was in use. The OCR has real limits:
- English only. Only an English language model is installed.
- Best on clear, horizontal, high-contrast text. Skewed scans, small print, unusual fonts, handwriting and low contrast give more mistakes.
- It can misread. Each recognised region shows a confidence figure, and you are told to check every region before you apply a change.
- One page at a time. In the PDF editor you run OCR on the page you are on, not on the whole document at once, and a large scan can be slow on an older device.
- It is an editor, not an extractor. It does not produce a searchable PDF of the whole document or a text file of all pages.
Which situation are you in?
| What you see | What it is | What to use |
|---|---|---|
| You can select and copy text in a PDF viewer | Real text layer | PDF to Text |
| The page looks like a photo of paper, and you cannot select text | Scanned image | OCR in the PDF Text Editor |
| You have a picture (JPG, PNG, WebP) of text | Image | OCR in the Image Text Editor |
| You are not sure | Unknown | PDF Text Density Checker |
The Text Density Checker measures how much extractable text each page has and reports whether the document is text-based, scanned or a mix, so you do not have to guess by eye.
A document can be both
Many PDFs mix the two, such as a digital cover followed by scanned attachments. Extraction then gives real text for the digital pages and little or nothing for the scans, which is what a "mixed" result from the checker is telling you.
Related tools
Check the page count and other facts with PDF Information. Pull out just the pages you need first with Extract PDF Pages (see PDF page ranges explained), and use PDF Image Extractor if what you want is the pictures inside a PDF rather than text.