Scanned PDF or EPUB: which to use
For a scanned book, a cleaned PDF is usually the safer choice. EPUB lets you change the font, but getting there means rebuilding every page from OCR, and footnotes, page numbers and whole lines often go missing.
Key takeaways
- An EPUB is text that flows to fit any screen. A scanned PDF is a picture of each page. They aren't two versions of the same file.
- To make an EPUB from a scan, software first has to read every word with OCR, then guess where paragraphs, headings and footnotes begin and end.
- Calibre's own manual says complex, multi-column and image-based PDFs aren't supported by its PDF conversion.
- Pick EPUB for a plain novel with clean print. Pick a cleaned PDF for anything with footnotes, tables, pictures, old type or page references.
Two different kinds of file
People often talk about a scanned PDF and an EPUB as if one could be saved as the other. They’re built differently.
- An EPUB holds text, with marks for headings, paragraphs and pictures. Your e-reader lays it out fresh for its screen and your chosen font size. There are no fixed pages.
- A scanned PDF holds one image per page, exactly as the scanner or phone saw it. There may be a hidden text layer from OCR, or there may be nothing but pictures.
Going from the second to the first isn’t a format change. It’s a rebuild of the book from scratch, from whatever a machine can read off the page images.
What happens in a conversion
A converter has to do four jobs, and each one can go wrong on a scanned book:
- Read the words. OCR turns each page image into text. Faint print, grey paper, curved lines near the spine and old typefaces all cause misreads.
- Throw away what isn’t the text. Page numbers, running heads and printer’s marks need recognising and removing. Miss them and they land mid-sentence.
- Put the paragraphs back together. Every line on a printed page ends with a break. The converter has to decide which breaks end a paragraph and which just end a line, and rejoin words hyphenated across lines.
- Decide where everything else goes. Footnotes, captions, tables, sidebars and two-column pages have no obvious place in a single stream of text.
Calibre’s manual is frank about this. Its PDF conversion relies on the text already inside the PDF, which it warns can be very different from what the page shows, and it doesn’t support complex, multi-column or image-based documents.
What tends to go missing
Readers who convert scans to EPUB run into the same handful of faults:
| What you lose | Why it matters |
|---|---|
| Whole lines or passages | A line the OCR couldn’t read simply isn’t there, and nothing tells you |
| Footnotes in place | They get merged into the paragraph or piled at the chapter’s end |
| Original page numbers | You can’t cite the book or use its index |
| Tables and verse | Columns and line breaks collapse into run-on text |
| Pictures with captions | Images move or drop out, and captions drift away from them |
| Old letters and mixed scripts | Fraktur, Greek or long s turn into nonsense characters |
None of these show up on a quick flick through. You find them on page 140, when an argument jumps or a footnote marker leads nowhere.
The middle option: reflowing the page
k2pdfopt takes a different approach. It finds the regions of text on each page image and reassembles them into new, smaller PDF pages sized for an e-reader, without needing OCR to be perfect. It’s free and can work well on single-column text. The catch is the same one as EPUB: pages are cropped, reflowed and split, so the page numbers in the output no longer match the printed book.
Which to choose
Best suited for: deciding before you spend an evening on a conversion.
| Your book | Choose | Why |
|---|---|---|
| A modern novel scanned cleanly, single column | EPUB can work | Few footnotes, nothing to cite, and you gain font control |
| A textbook, study or reference book | Cleaned PDF | Page numbers, figures and footnotes must stay put |
| An old or public-domain book on yellowed paper | Cleaned PDF | Faded type and old letterforms trip up OCR |
| Poetry, plays, music or maths | Cleaned PDF | Line breaks and layout carry the meaning |
| A book in two columns on a 6-inch screen | Try k2pdfopt, or read in landscape | Neither EPUB nor a full-page PDF suits small screens here |
Making the PDF read like a book
The usual complaint about the PDF route is the tiny text and grey pages. That’s the file, not the format, and it can be fixed:
- Crop the margins once for the whole book, so the text fills the screen instead of the white border round it.
- Turn grey paper white and print black, keeping pictures in greyscale.
- Add an OCR text layer under every page, so search, highlights and the dictionary work, but nothing you see on the page is replaced by machine-read text.
FixMyScan does all three in one upload, for books up to 1,000 pages. If a word is misread, the only effect is on search; the page you read is still the scan, cleaned.
When EPUB really is better
If you need large print far beyond what a cropped page can give you, or you listen with text-to-speech, reflowed text has a real advantage. In that case, check the converted EPUB against the scan for at least one chapter, and keep the PDF for reference.
Questions
Can I convert a scanned PDF to EPUB?
Yes, but the result depends on the OCR. The converter reads the words off each page image and reflows them. Every misread letter, merged footnote and broken paragraph ends up in the EPUB, and the original page numbers usually go.
Is PDF or EPUB better for an e-reader?
EPUB is better for books that began as text, because you can change the font and the words rewrap. For a scanned book, a PDF cropped and cleaned for the screen usually reads better and keeps every page intact.
Why does my converted EPUB have random page numbers in the middle of sentences?
The page numbers and running heads were part of the scanned page. A converter that doesn't recognise them pours them into the text along with everything else.
Does Calibre convert scanned PDFs?
Calibre uses whatever text is already in the PDF. Its manual warns that this can differ greatly from the page you see, and that image-based and multi-column PDFs aren't supported.
Does Kindle accept EPUB?
Amazon's Send to Kindle page lists EPUB among the file types it accepts, along with PDF.
Is PDF or EPUB better for Kobo?
Kobo opens both. Kobo's help lists EPUB and PDF for its eReaders. The same rule applies: EPUB for born-digital books, a cleaned PDF for scans.
Can FixMyScan make an EPUB?
No. It keeps the book as a PDF with every page where it was, and makes that PDF readable: margins cropped, paper white, text black and searchable.