A searchable PDF from a photo of a document
Updated October 31, 2026
You photographed the minutes of a meeting, a page of a textbook or a letter, and now you want to find a name in it, copy a paragraph or have your computer's search find the file by its contents. A photo cannot do any of that, and neither can a PDF made from a photo: it is still a picture. What you need is a PDF with a text layer: the picture of the page, with the recognised words laid invisibly over it at the right places.
Scan to PDF makes that file on your device, with optical character recognition (OCR) running in the browser. The steps below also cover what makes the recognition good or bad, and how to verify it.
1. Get a clean page first
Recognition is only as good as the picture. Open Scan to PDF, photograph the page or choose the photo from the gallery, and check the result: straight, evenly lit, text horizontal. Fix a missed corner with the magnifier. Choose the filter with the engine in mind: Black & white gives clean letters on plain text; Natural or Grayscale is safer for pages with faded print, where thresholding can thin the strokes. Look at the smallest print in the Before / after view; if you cannot read it, the engine cannot either.
Turn pages the right way up: Check pages proposes a rotation for pages that look turned, and a page scanned sideways gives nonsense text.
2. Choose the language
Tick Searchable text (OCR) and pick the language of the document, not of your phone: the engine knows the vocabulary and letter shapes of that language, and accuracy falls sharply with the wrong one. Fifteen languages are available: English, Romanian, Spanish, Portuguese, French, German, Italian, Dutch, Polish, Hungarian, Turkish, Czech, Russian, Ukrainian and Greek. A document that mixes two, such as an invoice with English terms, can carry a second language. The language data is fetched from this site once and kept on the device.
3. Save and verify
Tap Save PDF. The words of each page are recognised on the device and written into the PDF as invisible text under the picture; the picture itself is not changed. Then verify: open the PDF in any reader, search for a word you can see on the page, and check that it is highlighted in the right place. Select a line and copy it into a note. If the search finds nothing, the page was probably too dim or in the wrong language; fix it and save again.
Desktop search on Windows and macOS can index the text too, which is the point for anything you will want to find again.
Text as a file instead
When you want the text itself, to paste into a document or to translate, run the photo or the saved PDF through Image to text (OCR): it gives the text with a marker between pages, to copy or to save as .txt. The searchable PDF and the text file come from the same engine.
Limits to know
Printed text only: handwriting is not recognised. Columns, tables and footnotes come out in reading order as plain text, not as a layout. The text layer is not editable inside the PDF; it is there for search and copy. Accuracy is high on a clean, straight page in the right language and drops on small print, dim or blurred photos and wrong languages; the only honest check is a search for a word. Recognition takes a few seconds per page on a phone and can be cancelled. Redacted areas are removed from the text layer as well as from the picture.
Tips
- Scan at Balanced or Maximum quality when the print is small; the Small preset is for sending, not for recognition.
- Photograph a page at a time: two facing pages in one picture give an interleaved reading order.
- A second language helps with mixed text but slows the recognition a little; leave it on None for a single-language page.
- Verify with a word that is rare on the page, such as a name or a number, rather than a common word.
- The saved scan stays on the device: you can come back, change the filter of a page and save the PDF again with better text.
Questions
- Is the text sent to a server?
- No. The engine (Tesseract) and the language data run in a Web Worker in your browser; the pages never leave the device.
- Does the searchable PDF look different?
- No. It looks exactly like the scan; the text is invisible and only shows when you search or select.
- Can I make an existing scanned PDF searchable?
- Yes: Image to text (OCR) takes a PDF and offers a searchable PDF as output.