Not sure what you mean by comparing with screensavers

If you want to create a snapshot, you'll necessarily have to :
- grab the resulting HTML output
- use this output to create a picture. It could range from using an existing
borwser to grab the picture to creating your own rendering engine (of course
the formaer option is likely much easiert than the later)
- of course it will be time consuming if you want to perform this on all
those requests...

The whole picture may help. For example what kind of analysis do you need to
perform ? For now I don't understand why you need a picture of the rendered
image. Wouldn't it be enough to keep the HTML content as you can both
analyze the content (and much more easily depending what kind of analysis
this is) and to render this directly in whatever browser the user who needs
to get back at those data uses, rather than to present him a snapshot..