(This post has been moved.)
Silly me. I’m an old programmer and I pride myself on trying to get my ebooks “just so”, as if I were writing a piece of code. I want to create worthy offerings to add to humanity’s river of books; at the very least, they should be shiny and well-scrubbed.
So when Jaye Manus offered to judge the formatting of a few books from her blog fans, I hopped right on board, and she was very kind in her review. But I read with horror things like “squishy line spacing” and links to chapters not working quite as they should, systematically.
I use an EPUB reader and hadn’t seen the book on a Kindle device other than the PC Previewer, so it was useful to see this from the Kindle reader’s perspective, since none of my buyers had complained (yet). Without a Kindle device, I hadn’t realized quite how irritating it was to not properly trigger the “Cover” and “TOC” hard buttons.
Now on the one hand, it wasn’t really broken, but on the other hand, I want perfection in book formatting, and some cosmetic and graphic flourishes. I’m not willing to settle for “good enough”, so Jaye was nice enough to coach me through some of the issues.
If you’re content with auto-conversion from EPUB to MOBI or vice versa, or output direct to ebook formats from products like Scrivener, then this is overkill for you and you can stop reading now. But if you want as much control as possible over the results without killing yourself, you might find the following approach useful.
Originally I formatted my ebooks in raw HTML using tips from a variety of online recommendations, like Guido Henkel’s. The first book was a real learning curve for me, but after that it wasn’t difficult to just do the same for later books. Automated search/replace macros took care of things like wrapping lines with tags, and so forth.
I took the HTML output, opened it in Calibre, added the cover and converted it to EPUB using Calibre defaults (more or less). I did the same with a separate conversion to MOBI, which required me to maintain a different HTML file because of the way Calibre generates the MOBI TOC. These outputs were what I was uploading to the distributors. The MOBI files produced this way were not ideal, possibly because of AZW vs MOBI choices (in other words, me as a Calibre user, not necessarily Calibre as a tool), and I was left with two files to maintain (EPUB and MOBI) and the Smashwords EPUB as a third file, since my approach wasn’t modular. So every time I found a typo…
Jaye has taught me a better way…
I poured my HTML file for a book, broken up into chapters, into a Sigil template. Each part of my book has a separate file: Beginning Blurb, Title Page, Copyright Page, Also-By-This-Author Page, Small TOC, Chap 1,…, Chap Last, Guide & Name Index, If-You-Liked-This-Book Page, Excerpt-From-Next-Book Page, Author Bio Page, Long TOC.
I have 3 outputs: MOBI, EPUB, Smashwords EPUB. The difference between Mobi and the two EPUBs is that the “stylesheet.css” file is a little different between MOBI and EPUB, and the Cover page is treated differently (EPUBs require an extra step). The difference between my EPUB and my Smashwords EPUB version is that the Copyright Page has different content.
Now, thinking in the long term, I expect that the differences between MOBI output and EPUB output are likely to be persistent, and other devices may come along and generate different optimum stylesheet requirements. So I’m fine with having two different (but very similar) stylesheets which I maintain externally and copy in as needed into the stylesheet.css shell.
Likewise, the fact that Smashwords requires its own ISBN means only that I maintain two different external Copyright Page HTML documents and copy the contents of whichever one I want into the shell in Sigil.
Both of these make use of a simple modular structure.
So, what happens when I finish a new book and want to format it?
1) I copy the Sigil file (MOBI version) from the previous book and rename it.
2) I create Copyright.HTML files for both the normal and Smashwords copyright pages by copying the ones for the last book, renaming them, and updating the content.
3) I create a new Title page (it’s a graphic) and a new Cover.
MAIN CONTENT (MOBI)
4) I work on the MOBI version first (it’s the master). I copy the text in, chapter by chapter, the front blurb, and the back excerpt. I run saved searches to wrap the lines with <p></p> tags and to convert special characters to named entities.
5) I update the Also By and If You Like This Book pages by hand.
6) I run a saved search to update the Title field on all the HTML pages to the new work, and update the equivalent fields in the TOC.NCX and CONTENT.OPF files.
7) If the book is a little longer or shorter (number of chapters) than the last one, I update the TOC.NCX and CONTENT.OPF files and the HTMLTOC file.
8) I update the metadata in the TOC.NCX and CONTENT.OPF files. This allows me to do some things that either Calibre doesn’t, or I don’t know how to find, such as set a UUID (Unique User ID) for my short stories that don’t have ISBNs, embed book descriptions, add keywords, etc. There’s a great tool for this that Jaye told me about:
9) Run the file through Kindle Previewer (which runs Kindlegen) and check the results.
How much time does this take? I just updated my entire backlist (3 novels, 5 short stories, 1 story collection) to Sigil – it takes me about an hour for a novel, and 20 minutes for a short story. Making the short story collection from the already-formatted short story files was truly trivial.
10) Substitute the content of the EPUB stylesheet into the stylesheet.css.
11) Run the Sigil tool to insert the cover. (Kindlegen does that a different way for MOBI).
12) For the Smashwords variant, substitute the contents of the Smashwords copyright.html for the default one.
That’s it. I import the files into Calibre for one last look to make sure they seem healthy, and do a quick scroll through on my EPUB reader. If I find typos, I fix them in the MOBI version (and the Scrivener original) and redo steps (10)-(12).
Why not use Calibre? I am confused by the various options and clearly, for MOBI conversion, I wasn’t doing it quite right. Also, my original HTML file was one big file with a stylesheet and all chapters together, making modular changes clumsy. Calibre created its own version of the styles it found, and they weren’t always what I expected. It’s a big black box to me, and there were some issues with the results, which may be my fault, not Calibre’s.
Why not use Scrivener? Like Calibre, you are at the mercy of whatever Scrivener decides to do to instantiate the different conversions. Since the Scrivener text isn’t in HTML, there are all the issues of named entity conversions to deal with, and you have little control over the default styles. The results may be clean, but you can’t do anything special, such as use graphic chapter heads, scene dividers, and so forth, at least not in the Windows version. Perhaps there’s a way…
Other tools, like Scrivener, will take your word processor input and generate EPUB and MOBI output, but the black box in between what you write and what they produce leaves you at the mercy of the limitations of others, and so your output will remain at best functional vanilla. That’s not a bad thing, but we can do better.
It’s really not that hard to go through the learning curve once. After that, each new book becomes quite easy. Your book designers or people like Jaye can help you get started by setting up the first one and explaining how it works.