14

If I have an option of reading e-books in a range of formats (ebub, mobi, Sony LRF, .lit, rtf, MS Word's .docx, Rocket, Amazon's Kindle format, pdf), which one takes the least amount of space over the range of different book structures?

Assume that the book doesn't contain any advanced features that would take a lot of space in one format yet be incompatible with other formats. In other words, an edge case of "well, TXT file is smallest" isn't really applicable.

Please note that any answer should be based on either someone actually having compared identical ebook sizes in different formats and published the findings; not guesses or personal experience comparing 1-2 books.

Improve this question
9
  • Is there (or do you have) an example set for the range of different book structures?
    – Anthon
    Dec 22, 2013 at 20:59
  • @Anthon - do you mean an example of format list? I gave one in the first paragraph. Or do you mean actual book in those formats? If so, I can provide one for some formats (Baen Free Library has ~50-70% of them for same books) but as I said, single book is a VERY bad basis for an answer since different ebooks have different content that takes different space (more markup, more pictures, more structure, etc...)
    – DVK
    Dec 22, 2013 at 21:07
  • I read 'the range of different book structures' as Novels vs. scientific articles, vs children books etc. i.e. books with different statistics about wording and grammatical constructs. As those different structures might have different compression statistics in the different (storage) formats.
    – Anthon
    Dec 22, 2013 at 21:48
  • @Anthon - Yes, that's what I meant, though without the implication that the syntactic structure was necessarily driven by semantic one. Simply that some books have different amount of markup which takes space as well as different compressability.
    – DVK
    Dec 22, 2013 at 21:51
  • But within one format (e.g EPUB), some files will have higher amount of tagging. I was wondering if a set of files in one format, with differing characteristcs already exists. If so, you would only have to convert them to the other formats (using e.g. ebook-convert) and record their sizes to get a comparison table.
    – Anthon
    Dec 22, 2013 at 21:57

2 Answers 2

Reset to default
13

This is something I find interesting. I'm going to post an answer with some findings. As I discover more, I'll expand on my answer.

Old, But Still Useful Comparisons

This document is fairly old (warning, link is a zip file containing a pdf) and the information may not accurately reflect how things are today. It also does not contain many of the popular formats of today (there is none of the epub formats listed and mobipocket (prc) I believe as a predecessor to .mobi). I'll list the details for historical purposes though.

Excerpt from the report (Copyright 2002 by ePMA -- ePublishing Marketing Associates -- www.epmaasoc.com):

To make this comparison, we took the Project Gutenberg plain text file of the book The Metal Monster by Abraham Merritt, stripped it of the PG legal Disclaimer and such. Next we removed the linefeed/carriage returns from the end of each line, so paragraphs would flow for different formatting, and then added basic markup with HTML, to provide paragraph breaks and chapter headings. Except where noted, this HTML file was then used to create the eBook in other formats. no cover or other images were included.

NOTE: This was done for a non-illustrated volume. There was talk of doing another test in the future with an illustrated volume, but I was not able to find it.

Here were the findings:

Format                               Kbytes% Txt
Rocket (fb)                        248K   52%  
Mobipocket (prc)                285K   60%  
MS Ebook (lit)                    292K   61%  
Plain text (txt)                    477K   100%
Web Document (HTML)     487K   102%
Embiid (ubk)                       489K   103%
MS Word 2002 (doc)         620K   130%
hiebook (kml)                     654K   137%
Adobe Acrobat (pdf)        691K   145%
MS Wordpad (doc)            952K  200%
eBook Pack Express (exe)2860K600%

UPDATE

This is a comparison of various ebooks I own in multiple formats. I did not create or convert the formats personally and I am unsure of the exact versions (e.g. epub2 vs epub3). I still thought this might be useful information since all books were generated by the same company (manning). These are all programming books, and as such, they have some images in them and embedded markup language. This should provide a decent comparison vs. something that is only text or mostly images.

The following list of books are compared in epub, mobi and pdf (sizes are in KiloBytes, smallest bolded).

Findings

Average Sizes

  • epub - 8,396K
  • mobi - 7,717K
  • pdf - 11,426K

In almost all cases, the epub was the smallest file. Mobi was slightly larger and pdf was the largest. However, the average size of the mobi files were actually smaller overall. This was due to one very large epub file for the HTML5 for .Net Developers book. This book skewed the results since the file was nearly twice the size of the other formats. My best guess is because this file contains many HTML code examples, which may cause problems for the epub format (basically XHTML). Perhaps this is an epub3 format, which supports HTML5 (the book title is a good hint). Again, I did not generate these files, so I am not certain of the reason.

To combat this single anomaly, I decided to remove that book from the averages. When removing the book, the averages of the remaining books looks like so:

Average Sizes (removing HTML5 for .Net Developers - i.e. the anomaly)

  • epub - 5,703K
  • mobi - 6,678K
  • pdf - 11,134K

Take it for what it is worth, but in these findings it suggests that the epub and mobi formats are quite close in size, while the pdf format is nearly double.

Answer to be continued with newer formats and more research.

Improve this answer
0

for a comparison of AZW3 with epub, using a sample of 10 science fiction novels converted from AZW3 to epub by Calibre, a 555K AZW3 is 365K as an epub. so using ratio = (((file1Size - file2Size) / file1Size) * 100), I get 555 - 365 / 555 *100 = 34.23% with the .epub being 34% smaller. all 10 I tried were equivalent with variations of less than 1%

these are kindle books that have had drm removed and been converted using Calibre plugin for DeDRM and Calibre's convert from AZW3 to EPUB at default settings. The goal was to archive and convert my (Purchased) Kindle library to epub so I could read them in fbreader, my preferred android ebook app. BTW, converting my entire library to epub and deleting the other formats, mostly AZW and MOBI, took the size from 24G to 17.2G, a 33.0097% reduction in size on disk :)

Improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.