Inksky
You’ve got a document — maybe a Word manuscript, a scanned PDF, a saved web page, or a raw TXT file — and you want to read it on your e-reader. Here’s what works, what doesn’t, and how to get the best results from each conversion type.
Word (DOCX) → EPUB: the easy one
DOCX to EPUB is the smoothest conversion path. Microsoft Word’s own “Export to EPUB” (File → Export → Create EPUB) produces clean, reflowable EPUBs with hierarchical headings, embedded images, and basic metadata. Calibre matches or exceeds this quality with more control over the output.
Best practices:
- Use Word’s built-in heading styles (Heading 1, Heading 2) for chapter titles. These become the EPUB’s table of contents automatically.
- Avoid text boxes and WordArt — they don’t survive conversion.
- Resize images to 1200–1600px wide before inserting. Calibre won’t downscale them for you.
- Remove tracked changes and comments before converting. Every revision marker becomes visible clutter in the EPUB.
Result quality: Excellent. A properly styled DOCX converts to a clean EPUB with correct chapter breaks, embedded fonts, and working internal links.
PDF → EPUB: the hard one
PDF to EPUB is the worst conversion in the ebook world. PDFs are designed for print — fixed pages, absolute positioning, no concept of paragraphs or chapters. Converting one to a reflowable EPUB is like asking Photoshop to output clean HTML. It can be done, but expect to spend time cleaning up.
What goes wrong: Line breaks become hard returns after every sentence. Headers and footers get mixed into body text. Tables break. Hyphenated words at line endings become permanently hyphenated. Multi-column layouts become a shuffled mess of text blocks.
When it works: Simple, single-column PDFs (novels, reports) convert acceptably with Calibre’s heuristic processing. Books with complex layouts (textbooks, magazines) are better left as PDFs and read on a large-screen device like a Kindle Scribe or reMarkable.
Improving results:
- In Calibre, enable heuristic processing (Convert → Heuristic Processing → check “Enable heuristic processing”).
- Use the “Unwrap lines” option with a factor of 0.4–0.5 to merge broken lines into paragraphs.
- For scanned PDFs, run OCR first (Calibre can’t convert image-only PDFs to text).
Better than Calibre for PDFs: k2pdfopt (free, open-source) reformats PDFs for small screens without full conversion, preserving layout while making text readable. It’s not EPUB output, but for PDF reading on a 6-inch device, it’s the most practical solution.
HTML → EPUB: straightforward
HTML converts to EPUB cleanly because EPUB is essentially HTML in a ZIP wrapper. Calibre handles HTML input well, preserving links, images, and CSS styling. If you’ve saved a web page or an article, HTML → EPUB is the recommended path.
Tips:
- Use browser reader mode to strip navigation, ads, and sidebars before saving the HTML.
- For web articles, Calibre’s “Fetch News” feature can pull content directly from RSS feeds.
- If you have multiple HTML files (one per chapter), Calibre can combine them into a single EPUB during conversion.
Result quality: Very good. HTML is EPUB’s native language.
TXT → EPUB: bare-bones but reliable
Plain text files convert easily but require manual structuring. Calibre will wrap the entire TXT in a single chapter unless you tell it where chapters begin.
Getting chapters: In the conversion dialog, go to “Structure Detection” and set “Detect chapters at” to a pattern like //* (double-slash followed by text) or use a regex pattern matching chapter headings (Chapter \d+). You’ll need to ensure your TXT file has consistent chapter markers.
Formatting: TXT has no italics, bold, or images. If you want formatting, add basic Markdown-like markers before your chapters and use regex in Calibre to convert them.
Result quality: Functional but plain. Good for novels and text-only content; useless for anything with formatting or images.
EPUB → Kindle format (AZW3/KFX)
Once you have an EPUB, converting to Kindle formats is straightforward with Calibre and the KFX Output plugin, or just use Send-to-Kindle (which converts EPUB automatically). See our KFX vs AZW3 guide for the trade-offs between those formats.
Conversion quality at a glance
| Source | Result | Effort | Recommendation |
|---|---|---|---|
| DOCX | Excellent | Low | Calibre or Word export |
| HTML | Excellent | Low | Calibre |
| TXT | Good | Medium | Calibre with structure detection |
| RTF | Good | Low | Calibre |
| PDF (text) | Poor to fair | High | k2pdfopt or read as-is |
| PDF (scanned) | Very poor | Very high | OCR first, then convert |
| EPUB → AZW3 | Excellent | Low | Calibre |
The bottom line
The conversion itself is rarely the bottleneck. It’s the source formatting that determines the result. A clean DOCX with proper heading styles converts beautifully. A messy PDF with multi-column layouts and scanned images will make you want to throw your computer out the window. Start with the best source file you can find, and you’ll rarely need to spend time cleaning up the output.