OCR stands for Optical Character Recognition. It's the technology that looks at a picture of text — a photo, a screenshot, or a scanned page — and figures out the actual letters and words, giving you real text you can copy, search and edit.
Without OCR, a screenshot of a paragraph is just pixels: you can see the words but you can't select them. OCR extracts the text so the content becomes usable.
You can extract text from an image right in your browser — nothing is uploaded.
How OCR works
Optical Character Recognition converts pictures of text into actual text. The engine first cleans up the image and finds the regions containing writing, then works out where lines and individual characters begin and end, then classifies each character shape, and finally checks the result against a language model so that context can correct visually ambiguous cases. That last stage is why choosing the right language matters so much: recognising Romanian text with an English model produces markedly worse output, because the dictionary used to resolve ambiguity is wrong.
What it handles well and badly
Clean, high-contrast, roughly horizontal printed text is recognised accurately — scanned documents, screenshots, printed pages, book pages. Accuracy falls sharply with faint photocopies, skewed or curved pages, decorative fonts, low-resolution photographs, and text over a busy background. Handwriting is largely out of reach for engines built on printed text; neat block capitals sometimes come through, cursive generally does not.
Getting a better result
The input matters more than any setting. Photograph or scan straight on rather than at an angle, get even lighting without shadow across the page, and use the highest resolution available — around 300 DPI is a common threshold below which accuracy degrades quickly. Crop to just the text before running recognition. Avoid photographing a screen where possible, since moiré interference confuses the engine badly.
Searchable PDFs
Running OCR over a scanned PDF does not replace the scan. It adds an invisible text layer positioned behind the page image, so the document looks exactly the same while becoming searchable, selectable and readable by screen readers. Because the original image stays on top, a recognition error never changes what a person sees — it only affects what a search finds.
Doing it privately
OCR is often run over passports, contracts, invoices and medical letters, which makes uploading them to a server an uncomfortable proposition. Running the recognition in your own browser means the document never leaves your device: slower than a datacentre GPU, but with nothing transmitted, retained or logged.