Converting (Kobo) epubs to reduce file size?

[Feather_Qwill]

Does anyone know of any method to reduce the bloated filesizes of Kobo epubs? I'm talking 2-3 MB per book (from Kobo), as opposed 250-350 KB (From other sources, such as OmniLit). I was thinking the Convert Book option would do it, but although I've been able to fix some of the other issues (ie, fonts that grow and shrink every couple of paragraphs), nothing I've done so far has significantly changed the file size.

Please help!

[theducks]

Originally Posted by Feather_Qwill

Does anyone know of any method to reduce the bloated filesizes of Kobo epubs? I'm talking 2-3 MB per book (from Kobo), as opposed 250-350 KB (From other sources, such as OmniLit). I was thinking the Convert Book option would do it, but although I've been able to fix some of the other issues (ie, fonts that grow and shrink every couple of paragraphs), nothing I've done so far has significantly changed the file size.

Please help!

Unpack the book and see where the TOTAL size is coming from.
THEN make a plan

[jackie_w]

It may well be because the html files in Kobo-style epubs (kepubs) are full of extra <span> tags, e.g. <span class="koboSpan" id="kobo.32.1">some paragraph text</span>. These wouldn't be removed by a calibre conversion as it is perfectly valid code. A long book may have many thousands of them, which adds up.

As theducks says, if you open one up you can see them. With a bit of fancy Regex search/replace all you may be able to remove them to turn the kepub back to a standard epub, but the risk of deleting actual book content makes this rather risky and I wouldn't recommend it.

Edit: The other likely suspects are embedded fonts and/or a lot of images.

[BobC]

Originally Posted by jackie_w

It may well be because the html files in Kobo-style epubs (kepubs) are full of extra <span> tags, e.g. <span class="koboSpan" id="kobo.32.1">some paragraph text</span>. These wouldn't be removed by a calibre conversion as it is perfectly valid code. A long book may have many thousands of them, which adds up.
Edit: The other likely suspects are embedded fonts and/or a lot of images.

The extra HTML markup will not contribute as much bloat as you might expect to the TOTAL book size. This is because of the mechanism used in the final zip compression of the component HTML files which tokenises commonly used "words".

A total size of 2-3Mb is likely to be, as mentioned, embedded fonts, assuming that both the Kobo and the OmniLit books both have the same images (or none at all).

BobC