Is your PDF a scan? How to tell, and what it changes

Every PDF looks the same in a viewer: pages with words on them. Under the surface there are two completely different kinds of file. One contains text: characters, in a font, at positions. The other contains pictures of text: one photograph per page, usually from a scanner or a phone camera. The viewer renders both identically, which is why the difference stays invisible right up until a tool behaves strangely, and then it explains a lot.

Three ten-second checks

1. Try to select a sentence. In any viewer, drag across a line of text. If individual words highlight, the file has real text. If your selection is a rectangle, or nothing selects at all, you are looking at an image.
2. Zoom to 400%. Real text stays razor sharp at any magnification, because it is redrawn from the font's outlines every time. A scan goes soft and pixelated, because there is nothing to redraw, only pixels to enlarge.
3. Check the size per page. A page of digital text costs a few kilobytes. A scanned page costs hundreds. A 40-page file at 300 KB is text; a 6-page file at 8 MB is almost certainly a scan.

Mixed files exist too, and are common in legal and administrative work: a typed contract with a scanned signature page stapled on, or a digital report with photographed receipts as an appendix. Each page answers the checks for itself.

Why every tool cares about the answer

Text extraction and conversion. Extracting text from a scan returns nothing, correctly: there are no characters to give. The same goes for PDF to Word, EPUB and the other converters, and for search inside any viewer. An empty result from a scan means the file holds no characters to extract.

Compression. The two kinds shrink for opposite reasons. A bloated digital PDF is usually carrying oversized embedded images, and compression re-encodes those while leaving the text untouched and sharp. A scan is one big image per page, so compression works on everything at once. Scans shrink dramatically as a result, but they also show quality loss sooner if you push the target too low, because there is no text layer to protect.

Editing. Text in a digital PDF can be found, covered and replaced. A scan has no text to edit; changing it means drawing on top of the image, which is what an editor's annotation tools do. Redaction differs between the two kinds as well. On a digital page, a black box drawn over a name leaves the text underneath for anyone who copies it out. On a scan, covering the pixels does remove what's visible, and the remaining risk sits in the metadata and any hidden layers.

Turning a scan into a text PDF

The bridge between the two kinds is OCR, optical character recognition: software reads the pixels, recognises the characters, and writes an invisible text layer behind the image. The page still looks exactly like the scan, but now it selects, searches, extracts and converts like a digital file. Sleek PDF's OCR does this in the browser, so the scan never leaves your machine, which matters most for the sensitive ones.

OCR output is only as good as the scan it reads. Straight, evenly lit, 300 DPI scans come through almost perfectly; skewed phone photos of crumpled paper produce guesses. For a document that matters, skim the extracted text before trusting it, and check numbers in particular: a misread digit is both the most common OCR error and the costliest.

If you get to choose, choose text

The easiest place to fix all of this is upstream, before the PDF exists. "Print to PDF" or "Export as PDF" from the original document produces real text; printing on paper and scanning it back produces an image and throws the characters away. The digital version is smaller, sharper, searchable and accessible to screen readers. Scan when the paper is all you have, and let the scanner's own OCR option run if it offers one; it is the cheapest moment to get the text back.