PDF vs EPUB

A PDF fixes the position of every character on a page of a specific size. That is the point of the format: a page designed for A4 looks identical on any screen, on any printer, forever. To achieve it, the file stores coordinates rather than sentences.

An EPUB stores the text and its structure, and lets the reading device decide where lines break. That is why font size, margins, line spacing and typeface are yours to change in an EPUB and not in a PDF.

On a large screen the difference barely matters. On a phone or an e-reader, a fixed A4 page has to be either shrunk until the text is too small to read, or shown at readable size with panning. Neither is reading.

Reflowing is the main thing, but several features on an e-reader only work on real text:

• Adjustable font size, typeface, margins and line spacing.
• Dictionary lookup and translation by tapping a word.
• Highlights and notes that survive a font size change, because they attach to text rather than to a screen position.
• Search across the book that returns text rather than page images.
• Text to speech, and the accessibility features that depend on a readable text layer.
• A table of contents the device menu can jump through.

Converting is a genuine loss of information, and it is worth being clear about what goes:

• Exact layout. Multi-column pages, sidebars, pull quotes and anything positioned relative to a page edge become a single flowing column.
• Page numbers. An EPUB has no fixed pages, so a citation to page 74 no longer has a target. This matters for anything you need to reference academically.
• Complex tables. A wide table that fit a landscape page rarely survives a narrow reflowed column intact.
• Equations and diagrams that were drawn as vector page content rather than embedded as images.
• Typography that was set by hand: kerning, deliberate line breaks in poetry, and hyphenation decisions.

Convert text-led documents: reports, long articles, roadmaps, novels, non-fiction without heavy figures, anything you intend to read straight through on a small screen.

Keep the PDF for reference material, anything with a layout that carries meaning, textbooks with figures and equations, sheet music, and anything you will cite by page number.

For a scanned document, neither format helps until the pages have been run through optical character recognition, because there is no text to reflow. A scan is a stack of images that happens to be shaped like a book.

Not all conversions are equal, and the difference is in how the tool reads the PDF. Text in a PDF is often positioned glyph by glyph with no notion of words, lines or paragraphs. A conversion that takes those glyphs literally produces text with words run together, lines broken mid-sentence, and no paragraph structure at all.

Doing better means reconstructing lines from glyph positions, inferring word boundaries from spacing, joining lines into paragraphs, detecting headings from font size and weight, and dropping running headers and footers that repeat on every page. That reconstruction is the actual work of a PDF to EPUB converter, and it is where they differ most.