Are there developpers out there, who (ab)use the CaptureScreenshot() function of their automated gui-tests to also create uptodate-screenshots for the userdocumentation?

Background: Whithin the lifetime of an application, its gui-elements are constantly changing.

It makes a lot of work to keep the userdocumentation uptodate, especially if the example data in the pictures should match the textual description.

If you already have automated bdd-gui-tests why not let them take screenshots at certain points?

I am currently playing with webapps in dotnet+specflow+selenium, but this topic also applies to other bdd-engines (JRuby-Cucumber, mspec, rspec, ...) and gui-test-Frameworks (WaitN, WaitR, MsWhite, ....)

Any experience, thoughts or url-links to this topic would be helpfull. How is the cost/benefit relation? Is it worth the efford? What are the Drawbacks?

See also:

Is it practical to retroactively write specifications documenting a system via automated acceptance tests?