Few steps to get this done after the environment set up. Basically prep html document, convert to xhtml, add to docx4j document, and generate.
1. (Preparing HTML) Wrap TinyMCE generated HTML in a root element such as div.
2. (Preparing HTML) Replace with
3. (Preparing HTML) Use regEx to remove height and width attributes from images. Affected version 2.8.1. See http://stackoverflow.com/questions/15659076/does-docx4j-convert-xhtml-to-docx-in-memory
4. (Convert to XHTML, Generate) Use XHTMLImporter to convert the html into xhtml. See http://priasardi.wordpress.com/2012/01/17/convert-html-to-docx-using-docx4j/
* Update – 2.8.1 seems to have trouble with underline and background color. 3.0.0 fixes that. Both versions have trouble with table border and table caption. 3.0.1 (unreleased) should fix table border. See http://www.docx4java.org/forums/pdf-output-f27/tabl-caption-and-border-issue-docx4j-3-0-0-t1763.html