Any idea about what backend technology might be in use for sites like photofunia.com and loonapix.com for image merging to create the effects? is it flash/flex or Open GL?
loonapix is doing server-side image processing to create the effect. If you look at the cloud over the ocean one, it looks like they just run a blur (perhaps a guassian blur) and remove the color through desaturation and then colorize it to blue, and then they overlay that on a stock image. This is a total guess, but it feels like they might have done this with Ruby On Rails -- if so, they probably use this: http://rmagick.rubyforge.org/
photofunia is also server-side. I also noticed that it uses a lot of face recognition to automatically place the face -- for that, they may be using OpenCV. Otherwise, it's mostly the same thing as loonapix, image processing and compositing on the server-side.
You could use many different image processing libraries to do that (ImageMagick or PIL). I work for a company that makes a .NET imaging SDK that can do it -- Atalasoft.
A few years ago, we posted this sample to show how to use blurs and noise generators to create random clouds. You'd need to do something like that except incorporate a photo into the process.