19 comments

[ 3.6 ms ] story [ 34.0 ms ] thread
I find myself taking screenshots of websites all the time. Pages that I find visually interesting, because I've had far too many experiences where I went back and it looks like a completely different website. Or sometimes to remember information that is shown on a website for only a short period of time, like holiday layouts and fun coming soon pages. And I even take screenshots of my own webpages so I can look back at it later.

I know there's archive.org, but they take too long to archive content when all I want is a screenshot of the recent past of a webpage.

So I created SnapRobot as a side project. It was built in less than a day.

You feed it a URL and leave it. SnapRobot monitors the page, taking screenshots whenever changes occur. You can come back whenever you want and relive pages of the past. :)

In the example, you can see how the top Hacker News items evolve over the course of a day.

Currently, any HTML change triggers a new screenshot so it works better for websites that don't dynamically generate different code on every request. No query strings for now.

This is pretty cool. Very well laid out and aesthetically pleasing for a 1 day project. Congrats!
That is pretty neat. Would it be giving it away if I asked how you generated the screenshots?
Thanks. There's a number of tools to generate screenshots, including browser automation (sometimes used for testing) and third-party APIs.

I'm using the WebThumb API right now: http://webthumb.bluga.net/home

If you want to build your own, you could use Watir or even shell scripting to put together a sequence that opens a browser, takes a screenshot, closes a browser, and crops the screenshot, etc.

At Sauce Labs, we use a combination of the Selenium API for taking screenshots and Castro for recording movies of browser pages. Things get fun when you can click button, type text into fields, or execute arbitrary JavaScript between screenshots.
Pretty neat and useful. May be you have already thought about it but I felt the need to forward/share that screen shot with my friends.
Thanks. Will look into making it easier to share - perhaps permalink short URLs?
Looks neat. You might need a threshold in changes made. Seems like most websites have small irrelevant things that change on every visit. Like number of comments on techcrunch, or a timestamp in the code, resulting in a screenshot on every try.
Thanks. Definitely that is a very important improvement to make.
I assumed the large numbers were dates, when they are actually times (13 = 1 pm, not the 13th). I'd either use 'am/pm' or add ':00' to each time to make this more apparent. Large numbers inside boxes is a similar pattern to calendars, which is why this can be confusing.
Right now they're actually just numbered in sequence (1, 2, 3). I have been playing around with it. Sorry for the confusion, I'll work on it!
This would be incredibly useful for publicly traded corps to help prove that they made information public at a specific time (i.e., disclosure requirements were met).
I am disappointed that entering http://snaprobot.com only gives an error page. I was hoping for an infinitely recursive snapshot.

Edit: tried it again, and I DID get a snapshot of snaprobot displaying the BBC site, which is its default page. I guess the error was an actual error (it was an indexing disallowed error). Kudos!