Compress EPUB ebooks in your browser
An ebook is heavy because of its pictures, not its prose. The cover and figures are re-encoded, the chapters and the table of contents are left exactly as they were.
or browse your device. Everything is processed on this device
No EPUB to hand?
How to compress an EPUB
It is read here in the tab and goes no further. Nothing about your library reaches a server.
72% at 1600px is more than an e-ink screen resolves, and typically halves an illustrated book.
It reports how many images it recompressed. On a novel it will tell you plainly that there was nothing to gain.
It is a valid .epub. Send it to a Kindle, a Kobo, Apple Books or Calibre exactly as you would the original.
What is actually inside an .epub
An .epub is a ZIP archive with a specific shape: one XHTML file per chapter, a stylesheet or two, an OPF file listing the lot in reading order, and a folder of images. Rename a copy to .zip and open it and every chapter is sitting there as a web page, because that is exactly what it is.
Which tells you where the bytes are. A novel is text, and text is astonishingly small: a 400 page book is roughly half a megabyte of characters, and the ZIP squeezes that to a fraction. If your ebook is 40MB, it is not the words. It is the cover art, the plate section, the diagrams in a technical book, or a scanned comic where every page is a photograph.
So this tool goes after the images, re-encodes each one as its own format at a quality you choose, and rebuilds the archive around them. Chapters, styles, fonts, footnotes, the table of contents and the reading order are all carried through unchanged.
That does mean a plain novel with no illustrations has almost nothing here to compress, and the honest result on one is a few per cent. The result panel will tell you that in as many words rather than dressing it up.
EPUB compression FAQ
How do I make an EPUB smaller for Kindle?
Send to Kindle caps attachments at 25MB, so turn on "Compress to a target size" and type 25000. Almost all of the weight in an oversized ebook is the cover and the figures, and a Kindle screen cannot show more detail than 1600 pixels of longest edge anyway, so the copy you read is indistinguishable from the one you started with.
Will my ebook still open on a Kobo or a Kindle after this?
Yes. There is one rule that decides this and we follow it exactly: the first entry in an EPUB archive has to be a part called mimetype, stored uncompressed, with no extra field on its header. That is what lets a reader identify the file from its first few dozen bytes. Plenty of generic ZIP tools break that rule when they repack an ebook, which is why a book run through one sometimes stops opening. This tool writes the archive itself and puts mimetype back exactly where it belongs.
Why did my novel barely shrink?
Because there was nothing in it to compress. Text is tiny and already deflated inside the archive: an entire 400 page novel is a fraction of the size of one photograph. If a book has no illustrations, the honest saving is a few per cent, and we would rather say so than hand you a placebo. The books that shrink are the ones with a plate section, technical diagrams or scanned pages.
Does it change the text, the chapters or the table of contents?
No. Every chapter file, the stylesheet, the fonts, the OPF and the navigation document are copied across unchanged, under their original names. Reading position, footnotes and internal links all still work, because none of them were touched.
What gets removed when I turn on "remove metadata"?
Calibre bookmarks and its added metadata tags, iTunes and macOS filesystem junk, and any duplicate cover thumbnail that nothing in the book references. The title, author, language, cover and the OPF itself are never dropped, because those are what stop the book showing up on your shelf as Unknown.
Can it convert an EPUB to MOBI or AZW3?
No, and it should not. A modern Kindle reads EPUB directly through Send to Kindle, and converting between ebook formats is a job for Calibre on a desktop. This page does one thing: it hands you back the same book, smaller.
What about a DRM protected book?
It cannot be read and is refused rather than mangled. A book bought from a store that applies DRM is an encrypted archive, and there is no password to ask you for. The same goes for a damaged file: we check every part against its own checksum and stop rather than quietly writing out something corrupt.
Is my ebook uploaded anywhere?
No. There is no upload endpoint in the code. The file is read and rebuilt in this tab, on your machine, and what you download is written straight out of memory.
Other compressors
Same format joins the queue. Anything else re-routes to its own tool.