SMIL, a sequencing/timing subset language of XML, will do just about anything you tell it to do.
What limits SMIL at this point in time is that, except for one or two very pricey and obscure development software tools, all the functionality must be implemented manually (read: hand-coded). When combined with Real Networks' Realpix transitions and RealText captioning, or with Microsoft's HTML+TIME hybrid timing language, some pretty amazing results can be produced.
However, the application with the greatest amount of flexibility and reach is Flash. Using Flash MX with Quicktime, or Flash MX 2004 with Quicktime, Real or WinMedia Player, video can be streamed. Whichever one you use depends more upon the bandwidth capability of your audience than anything else. In my estimation, Quicktime produces the highest quality video, but also the largest video files. Of the remaining two, in my own personal experience, the Real player is the more stable format, but has been known to come bundled with adware, spyware and tracking software in spite of the company's ongoing denial that it is doing so.
While SMIL seems poised to take center stage at the onset of the impending media convergence, at the moment, it's not quite ready for prime time.
If I'm not mistaken, I think Flash even supports SMIL to some degree, as the latest version has widened its support of XML overall. Talk about having your cake and eating it too!
Hope this info helps a little.