Why your converted PDF looks wrong — and how to fix it

An EPUB has no pages. Page breaks are invented by your reading app every time you open it, from your screen size and font settings. A PDF must commit to one set of pages permanently.

So a converter is not copying a layout across — it is creating one that never existed. Most "formatting problems" are really that new layout making a different decision from the one your reading app made. Knowing which symptom you have narrows the fix considerably.

The most common complaint by a wide margin. An image that sat beside its paragraph on your tablet now sits on the next page, or alone on a page of its own.

• The image is taller than the remaining space on the page, so it is pushed forward. A larger page size or narrower margins gives it room to stay put.
• The book uses floated images that need a specific text-column width. Landscape orientation often resolves this for wide figures.
• A fixed-layout book is being forced onto a differently-shaped page. Match the output page proportions to the book’s own.

A table with many columns was authored for a screen that could scroll sideways. A printed page cannot.

• Switch to landscape orientation, which is usually enough.
• Reduce the text size slightly — the table scales with it.
• Use a larger page size such as A4 or Legal.

• The book has no embedded fonts. The converter substitutes something similar; the text is correct but the look is not identical. Nothing can be done about this short of embedding a font yourself.
• The fonts were embedded but obfuscated. Publishers often scramble embedded fonts with a standard algorithm so they cannot be extracted. A converter that does not reverse it embeds noise, and the result is either substituted type or, in bad cases, garbled glyphs. This converter reverses both the IDPF and Adobe algorithms.
• The script needs fonts the renderer does not have. Gujarati, Hindi, Arabic, Hebrew, Chinese, Japanese and Korean text needs a font that covers those characters. If you see empty boxes, the renderer had no suitable font.

Almost always a problem in the book rather than the conversion. The package file lists a reading order, and only documents in that list are part of the book.

• A chapter marked as non-linear in the spine is excluded from the main reading flow by design — that is how endnotes and supplementary material are usually handled.
• A chapter file listed in the manifest but absent from the spine is not part of the book’s reading order at all.
• If chapters are missing in your reading app too, the EPUB itself is incomplete and needs re-downloading.

Usually deliberate. Books are often authored so chapters start on a right-hand page, which produces a blank left-hand page whenever the previous chapter ends on one.

Setting chapter breaks to continuous removes them, at the cost of chapters no longer starting on a fresh page.

A 2 MB EPUB becoming a 40 MB PDF is normal and not a bug. EPUB stores each image once, compressed; a PDF embeds every image at the resolution needed to print it.

• Lower the image quality setting. Balanced is usually indistinguishable on screen.
• Use the Compact preset, which combines smaller images with tighter spacing.
• If you are not printing, there is no benefit to keeping images at original quality.

If you cannot select text in the finished PDF, the converter rendered the pages as images. This is a real quality difference between converters, and it costs you searching, copying, accessibility and a great deal of file size.

A properly converted PDF contains real text. If yours does not, the conversion method is the problem, not the book.