If you don’t know much about graphical interpolation, be sure to check some great blog-posts that have come before mine (here and here, for example). After going over these posts, I was curious about the performance difference of these algorithms. I came across a page on the internet that compared some of the more popular interpolation techniques using images to provide concrete examples. See http://www.all-in-one.ee/~dersch/interpolator/interpolator.html. Basically, the author of this page rotated images 180 degrees (which involves loss, see the second aforementioned blog post), flipped the images (which does not involve loss) and compared the results to the original. You can see some nifty examples of how good the common interpolation techniques are. I like the linked resource in particular because it’s informative and straightforward (or, if you like to get more technical, the author even goes over some of the math towards the bottom of the page).

I’d be interested to see how this algorithm would fare against that test. Unfortunately, it looks like I’ll have to wait a while. A brand new algorithm from Jiri Patera and Armen Atoyan of the University of Montréal provides better quality and speed than techniques used hitherto. It seems that the details are under wraps for a while, at least until they get a few patents. The examples provided in this brief news post look extremely promising and impressive.