Low Quality when Compiling to Word or OpenOffice
Important: this bug should be fixed, so if you are experiencing this problem, the first thing you should try is upgrading your copy of Scrivener.
If you are compiling to PDF (using the "Proofing" print engine) .doc, .docx or .odt and notice a drop in formatting quality, or get a strange duplication of output files or folders when using accented characters in the full export path, then this article will help you resolve the problem.
There are three work-arounds to choose from:
First, consider using RTF if you can. In most cases, you will not need a binary .doc/x file. Rich Text is Scrivener's native format, and what is used to generate the .doc/x and .odt files in the first place. It is also broadly supported by nearly every word processor on all platforms. There are only a few situations where RTF is actually a worse option, or simply doesn't work (those who prefer Pages will find it has little or no support for RTF, and some mobile office suites also do not support RTF at all). As a side benefit, it will also compile dramatically faster since no third-party Java conversion is necessary.
If you do need a binary .doc/x though, then you'll need to first
make sure you have Java installed on your computer. The easiest way
to do this is to use the
Go/Utilities menu command in
Finder, double-click the Terminal icon, and copy and paste the
following, pressing Return after pasting:
If the result you get states something to the effect of: "command not found: java", then you will need to install it from the official Java website. After installation, run the test command again to ensure everything is installed correctly.
- Back in Scrivener, open the "Import/Export" Preferences
At the bottom of the window, enable Use system version of Java for conversion if installed.
This will tell Scrivener to ignore the embedded Java engine and use the system engine instead.
Try compiling again. If all goes well, you will get a clean .docx file this time.
Alternative to the above permanent fix, you can use the first option. Create an RTF and then use a word processor that is good at both RTF and DOC to make the final conversion. MS Word, Nisus Writer Pro and LibreOffice (free) are going to be the best, in that order. Whether to spend your time fixing the problem up front, or fixing it each time you compile will depend upon personal preference.
The accented character problem should be fixed in Scrivener 2.6, and extra detection has been added so that if the embedded engine fails for any reason whatsoever, it will first attempt to use any system Java installations, and only then if that fails will it drop to using the OS X converters.