Authoring in a markup based solution means giving up the idea of formatting as you write. Previously, formatting was by eye; with markup it will be by rule. Once a stylesheet is created, formatting by rule is faster and more consistent, but it does have
limitations.
Separate one dimensional components (linear text) form two dimensional components (graphics, charts, photos, complex tables). Automatic formatting engines (like an XSL-FO engine) process a string of elements (one dimensional) and flow them onto a page (two dimensional). Graphics are simply items in that sting of elements with a two dimensional size that must be published onto the a two-dimensional page space.
To achieve the benefits of automated formatting for XML-based content, graphics and their overlays must be treated as separate two dimensional objects, created outside of the main XML authoring system, and imported into it as graphics. Single sourcing screen shots and overlays to produce graphics that are imported into an XML based publishing system is a separate problem, and one that is not constrained by the XML system. All the XML publishing system knows is that it’s importing a graphic. How you create a graphic is up to you!
Moving to XML means fighting the monolithic thinking encouraged by DTP systems. An XML system can easily integrate elements from diverse sources. Don’t look for an XML solution with the capabilities of a DTP tool. Look for in an XML system that has the ability to integrate material from many sources. The system should provide solutions for individual problems and integrate those solutions into a working whole.
The solution to a graphic overlay problem may not be found in XML but the tool that solves the problem may use XML. With XML, find the best stand-alone solution a problem and to integrate the result into a publishing process.