Bringing translated novels to your e-reader: choose the right EPUB export, check broken TOCs, and tune margins for your screen
Tri Thu知秋
On this page
- How do the two EPUB options in the Download button differ?
- Bilingual EPUB vs. Bilingual TXT: which one should you pick to compare with the source?
- Blank pages or a missing table of contents: what should you check first?
- How to tune margins and line spacing for small screens (Xteink, mini readers, 6-inch)
- When should you embed a font to fix missing Vietnamese or Thai diacritics?
For the same finished translation, exporting from an original EPUB preserves its illustrations and chapter layout, while exporting from a TXT project builds a brand-new EPUB with screen-size layout options. If copying a book to your e-reader leaves you with blank pages, a broken table of contents, wide margins on a pocket screen, or missing Vietnamese and Thai diacritics, you can inspect the package structure and re-lay out the pages right in your browser before reading.

| Your situation | Which option to use | Key note |
|---|---|---|
| Novel translated from an EPUB file, original file still stored on the server | EPUB or Bilingual EPUB in the Download button | Writes the translation back over the original text positions, keeping images and most CSS; untranslated chapters stay in the source language |
| Novel translated from TXT, Markdown, or after the original EPUB was cleaned up after 30 days of inactivity | EPUB translation (rebuilt) in the Download button | Builds a new EPUB from translated chapters, omitting untranslated ones; offers 3 layouts or 5 device-screen presets |
| You already have a .txt or .md file on your device and want an EPUB with a table of contents | Convert TXT to EPUB with a table of contents tool | Runs in your browser without sending the file to a server; accepts up to 5,000,000 characters and 5,000 chapters |
| An EPUB opens to blank pages, is missing its table of contents, or seems corrupted | Check your EPUB before reading or translating tool | Checks package structure, OPF, reading order (spine), TOC links, and XML tags for files up to 50 MB; does not replace EPUBCheck |
| The book opens, but margins are too wide, lines are too loose on a small reader, or diacritics show as boxes | Optimize EPUBs for your e-reader screen tool | Resets margins and line spacing and embeds a .ttf/.otf font under 5 MB; reading apps may still override margins with their own settings |
How do the two EPUB options in the Download button differ?
In the "Download" menu, the export options depend on what source files are still stored for your novel. When a project starts from an EPUB file and that original file is still on the server, the menu shows "EPUB" and "Bilingual EPUB". Instead of generating a book from scratch, the system reopens the exact zip package you uploaded and replaces the text inside each paragraph, keeping illustrations, links, and most of the original CSS intact. Chapters that do not have a translation yet keep their source text in the exported EPUB. The full walkthrough for this route is in How to translate an EPUB novel into English and export a file you can read. Why a general chat window usually cannot repackage a whole EPUB with its table of contents intact is examined in Can ChatGPT read and translate a full EPUB novel? Output limits and how to keep the table of contents. How to keep the book title and chapter headings aligned with the exported table of contents is covered in Translating web novel book and chapter titles: keeping names aligned with the table of contents.
When a project starts from a TXT or Markdown file, or after the original EPUB file has been cleaned up following 30 days without translation or editing activity, the write-back options are replaced by "EPUB translation (rebuilt)" as soon as at least one chapter is translated. The rebuilt export packages an EPUB 3.3 file from your translated chapters, carries no original images, and skips untranslated chapters. Before building the file, the system also scans for and reports the number of characters invalid in XML that will be stripped. How chapter splitting and plain-text translation work is covered in How to translate a TXT novel into English and download it. If you have not started translating yet and want the full five-step workflow from file preparation to chapter review, read How to translate a whole web novel with AI: a 5-step workflow.
If you simply have a .txt or .md file on your device and want to turn it into an e-book with a clickable table of contents right away without signing up, open Convert TXT to EPUB with a table of contents. The tool runs on your device, accepts up to 5,000,000 characters and 5,000 chapters, lets you edit chapter titles, filter junk lines or duplicate chapter numbers, and attach a JPEG, PNG, or WebP cover up to 5 MB (WebP covers are converted to JPEG automatically so older e-readers can display them).
Bilingual EPUB vs. Bilingual TXT: which one should you pick to compare with the source?
The two bilingual formats arrange the original text and translation differently and pull the original text from two different places. In "Bilingual EPUB", each source paragraph is inserted right above its translated paragraph in the same book, marked with a gray left border, smaller type, and a dimmer color. Because that styling is attached directly to each paragraph, it still shows up on reading apps that ignore the book's stylesheet.
In "Bilingual TXT" (and similarly when exporting Markdown), each chapter places the entire translated chapter first and the original text below it. In addition, Bilingual TXT pulls its source text from the per-chapter copy stored in the app: once a chapter's source text is cleaned up when the novel is complete, that original half is gone and the downloaded file contains only the translation. By contrast, Bilingual EPUB reads the original text from the uploaded EPUB file itself, so it keeps both languages until the original EPUB file is removed after 30 days of inactivity.
Blank pages or a missing table of contents: what should you check first?
An EPUB is a zip package governed by strict packaging rules. Re-zipping a book with a regular archive utility can compress the mimetype file or move it away from the first entry in the archive, which causes many e-readers to refuse to open the book. Likewise, a single unclosed XML tag inside a chapter (xhtml-malformed) can leave a page blank or cut off a chapter halfway through, while a broken path inside nav or ncx pointing to a missing file makes table-of-contents entries unresponsive.
Before copying a book to your e-reader or uploading it for translation, drop it into Check your EPUB before reading or translating. The check runs in your browser (accepting files up to 50 MB and 20,000 internal package entries) and inspects the archive in order: container shell (mimetype, container.xml), book metadata (OPF and spine reading order), table of contents (nav, ncx), each XHTML chapter file, and UTF-8 encoding. Locations sharing the same root cause are grouped together so you can see which file and line failed.
One limit worth noting up front: this tool checks common structural issues only. It is not an official conformance validator, is not equivalent to EPUBCheck, and does not test how a specific device renders pages or images. If a book uses ZIP64 compression or contains encrypted resources, the tool explicitly reports that only part of the book could be checked instead of declaring the file error-free.
How to tune margins and line spacing for small screens (Xteink, mini readers, 6-inch)
Whether you export "EPUB translation (rebuilt)", build a book in the TXT-to-EPUB converter, or run Optimize EPUBs for your e-reader screen, you work with the same layout table. Alongside the 3 base choices for newly built books, "Standard", "Compact" (line spacing 1.3, paragraph spacing 0.3em), and "Spacious" (line spacing 1.6, paragraph spacing 0.9em), the "By reading device" group provides 5 presets tuned by screen size:
| Suggested preset | Page margin (px) | Line spacing / Paragraph indent / Paragraph spacing |
|---|---|---|
| Xteink X4 (4.3″) | 10 px | Line spacing 1.3 · paragraph indent 1.3em · paragraph spacing 0.3em |
| Xteink X3 (3.7″) | 6 px | Line spacing 1.25 · paragraph indent 1.2em · paragraph spacing 0.25em |
| Other mini readers (InkPalm 5, Boox Palma, Hisense…) | 12 px | Line spacing 1.32 · paragraph indent 1.3em · paragraph spacing 0.35em |
| Around 6″ (Kindle, Kobo, 6″ Android readers) | 16 px | Line spacing 1.45 · paragraph indent 1.3em · paragraph spacing 0.5em |
| Larger readers (Kindle, Kobo, large Android readers) | 22 px | Line spacing 1.5 · paragraph indent 1.4em · paragraph spacing 0.6em |
Two technical details help keep the exported layout from drifting. First, Chinese and Japanese web novels often start every paragraph with two full-width ideographic spaces (U+3000); when you pick one of the 5 device presets on a newly built EPUB, the builder strips those leading spaces automatically so they do not stack with CSS text-indent into a double indent. Second, page margins are applied through body margin alone (without @page), because readers that honor both, such as Adobe Digital Editions and Apple Books, would add them together and double the margin.
Even so, the numbers above are screen-size recommendations rather than measurements verified on every physical device (such as Xteink, InkPalm, Boox, Kindle, or Kobo). Many reading apps apply their own margin settings and override the file's margins: when I measured exported books in Calibre E-book viewer 9.13.0 on desktop, Calibre kept the left margin at roughly 20 to 21px for both Compact and Spacious builds, while only the line height changed from 22px to 27px in proportion to the CSS. Also, when running the EPUB Optimizer on an existing book, resetting paragraph spacing and indentation applies across text tags, so poetry or scene breaks may look different from the original. For the steps to transfer an EPUB onto a Kindle and check the package beforehand, see How to send a translated EPUB novel to Kindle and check for file errors before sending.
When should you embed a font to fix missing Vietnamese or Thai diacritics?
If a translation reads fine in your browser but shows square boxes, missing Vietnamese tone marks, or clipped Thai stacked vowels after you copy it to an e-reader, the usual cause is a built-in reader font that lacks those glyphs rather than corrupted character encoding. In my desktop test on Calibre 9.13.0 using a font with full diacritic coverage, the smallest ink gap between two consecutive lines stayed between 4px and 9px on Vietnamese text and 6px on Compact Thai text, so tone marks were neither touching nor clipped.
Under "Advanced options" in Optimize EPUBs for your e-reader screen, you can use "Embed your own font (optional)" and pick a real .ttf or .otf file under 5 MB. The tool verifies the font's magic bytes, embeds the file into the book package, and declares it on both body and p tags so books that set a separate font on individual paragraphs still pick up the new font. If you leave the font picker empty, the tool keeps the book's existing fonts; for Chinese, Japanese, or Thai books that already bundle a suitable font, leaving their original font untouched is the safer choice.
The EPUB Optimizer processes up to 20 files per batch (up to 20 MB and 5,000 internal entries per file, 100 MB total). After repackaging, it re-reads the new archive and verifies every entry so no chapters or images go missing before download. Books that contain encryption declarations, font obfuscation, digital signatures, rights metadata, ZIP64 headers, or UTF-16 content are refused up front rather than modified halfway.