Well done for misreading the argument, yet again.
All the integrations for doing it all use third party services for doing it. And all of those third party services set up dedicated resources for running browsers solely for gathering screenshots.
There never have been any mods for doing it purely on your server because it's beyond impractical to do it on your own server. And places like Google or BrowserShots have many dedicated machine resources for this purpose, far more than you can manage yourself conveniently or cheaply.
But hey, why would I know this? It's not like I've ever written HTML parsing code or anything to mimic browser behaviour. (That was irony.)
Should I get even more detailed as to the number of problems with implementing such things, or is this sufficient detail for you? Or do I need to go through and provide detailed research on every point I'm asserting because you're not willing to take my word?
ETA: OK, so let me explain some other details.
You are aware the web has malware on it, yet? You are also aware that there are a great many sites that use all sorts of JS and CSS trickery, yes? That means you have to completely clone a browser - or sandbox it - to be able to satisfactorily render the content on a lot of sites. Since you are doing that, you have to sandbox the browser very well because there is an awful lot of malware out there to protect against.
The alternative to avoiding the whole 'snapshot' aspect is to simply dump the content in an iframe which will cause the browser to load it for you automatically. And if there's a vulnerability in your browser that's exploited by the page being linked to, boom, you're hit even without you actually clicking the link. Yes, that's secure