Scanned PDFs and Text PDFs
Two PDFs can look the same on screen and be completely different inside. One is a description of text and shapes; the other is a photograph of paper. Almost every frustration people have with PDFs traces back to not knowing which one they are holding.
The two kinds of PDF
A text PDF, sometimes called a digital or native PDF, is what you get when software exports directly: a word processor, a spreadsheet, a design tool, a browser's print function. The file contains the actual characters, the fonts used to draw them, and vector instructions for lines and shapes. Nothing in it is a picture of a page.
A scanned PDF is a container around photographs. A scanner or a phone camera captures each sheet as a bitmap, and the PDF wraps those bitmaps one per page. From the file's point of view there is no text at all, only a grid of coloured dots that happens to look like writing to a human.
Mixed documents are common and are the most confusing case. A contract exported cleanly from a word processor, printed, signed, scanned, and then merged back with its appendix will contain both kinds of page in one file.
Telling them apart in a few seconds
The fastest test is to try to select a line of text with the mouse. If a neat selection follows the words, the page is text. If nothing highlights, or a blue rectangle covers the whole page at once, the page is an image.
Searching works equally well. Pick a word you can clearly see and search for it. A text PDF finds it immediately; a scan finds nothing, because there is nothing to find.
Zooming in far enough is the third test and the most definitive. Text drawn from fonts stays crisp at any magnification, because it is redrawn each time. A scanned page dissolves into visible pixels and, if it was compressed as JPEG, into soft halos around the letters.
Why scans are so much larger
A page of text stored as characters is extraordinarily compact. The letters themselves are a few kilobytes, and the font is stored once for the whole document no matter how many pages use it. A fifty-page report can easily be under a megabyte.
The same page scanned at 300 DPI is roughly 8.7 million pixels. Even compressed, that is commonly one to three megabytes per page in colour, and the cost repeats for every single page because no two photographs can share anything. This is why a scanned contract is often a hundred times larger than the document it was printed from.
Scanner settings make the difference enormous. The same sheet captured as black and white at 200 DPI might be 40 kilobytes, and as 600 DPI colour, 10 megabytes. Most office scanners default to something far higher than the document needs.
What OCR actually does
Optical character recognition reads the shapes in a scanned image and guesses the characters they represent. The usual result is a searchable PDF: the original photograph is left exactly as it is, and a layer of invisible text is placed on top of it, aligned with the visible words.
This is worth understanding, because it means OCR does not convert a scan into a text PDF. The page is still a picture, still the same size, and still just as blurry. What has changed is that searching and copying now work, by reaching through to the hidden layer underneath.
It also means OCR errors are invisible. A character misread as something else looks perfectly correct on screen, since what you see is the original photograph, but it will not be found by a search and will be wrong if copied. Recognition quality degrades quickly with low resolution, skew, handwriting, and unusual typefaces.
What this means when converting to images
For a text PDF, converting to an image is a genuine rendering step. The characters and vectors are drawn fresh at whatever resolution you ask for, so raising the scale genuinely adds detail and the output can be made as sharp as you like.
For a scanned PDF, there is no new detail to add. The page already is an image at a fixed resolution, so rendering it larger only enlarges the pixels that are already there. If a scan was captured at 150 DPI, exporting at 600 DPI produces a bigger, blurrier file and nothing else.
The practical rule is to match the scan rather than exceed it. Try a middle scale first, look at the result, and raise it only if the text is genuinely getting crisper. Once extra scale stops improving legibility, you have reached the resolution of the original and everything beyond it is wasted bytes.
Placing page images in slides and documents