Hi there @rolisize
Yes that post has been brought to my attention.
The details of how the stack works are at the bottom of the demo page, but I will add them below for clarity.
Suffice to say, some of the comments in that post are completely inaccurate, and disappointing.
SO HOW DOES CAFFEINE LOAD THE PAGE SO FAST?
Normally when your page contains a lot of large images, every single image has to be fully loaded into the browser before the page load is complete. Images can be extremely large in file size and extremely slow to load, especially when there are lots on the page.
What Caffeine does is makes a tiny copy of your image (just 50px) when you drop it into Rapidweaver. When the page initially loads it displays these tiny images stretched out to be the same size as the original for display.
Then once all those tiny images have loaded (just a few milliseconds, because they are tiny) the page will reach a loaded state. This is at least 20x faster than normal being conservative.
So what now. So you will briefly see those tiny images filling the space of the finished fully loaded large images, but they do give a good representation of whats about to appear.
Now Caffeine will load the full size images that it finds the user can actually see in the current window, (a second or so), and the user can now see any images that are visible in the window in all their glory.
What about the rest of images that are not yet visible in the window until the user scrolls down the page?
Well, as the user scrolls down the page, just before these other images come into view, the original image is loaded. The tiny blurry image fades out and the full size original fades in.
One other thing to note. Caffeine will never load a full size version of your image on a smaller screen. that would be a waste of bandwidth, and time. Instead, when Caffeine detects a smaller screen device it loads another smaller copy of your original image that is more suitable for this screen size. #stayfast