how does google create the instant preview images?

I was wondering how google is capturing all those websites that are featured in google's instant preview? I'm sure they are not using a thumbnail service (like www.thumbalizr.com, websnapr.com, snapcasa.com, thumbshots.com) but rather use their own software. BUT: given that google captures A LOT of websites, they must have a very sophisticated system. PLUS: this generates HUGE amounts of data (jpgs?). Does 开发者_Go百科somebody have more insight into how google does this?

---

Yes, it's something like that. Their webmaster pages hint that they render the page with the same engine Chrome uses, and the preview is based on the result.

---

It's hard to say, but here's some info from a Google project manager discussing it:

http://googleblog.blogspot.com/2010/11/beyond-instant-results-instant-previews.html

It says in part:

"we match your query with an index of the entire web, identify the relevant parts of each webpage, stitch them together and serve the resulting preview completely customized to your search—usually in under one-tenth of a second"

That plus looking at the source of a preview page suggests that they're using their own index (the same webcache.googleusercontent.com that is used to serve the Cached pages) to serve JPEG Base64 image strings as screenshots.