Rosa asked me a week ago when I would make something beautiful. She is involved with a lot of net art and I guess she referred to the fact that I do a lot with web scripting, crawling and scraping. However, as I am not a designer all things I make are ugly, unless a designer cooperates with me.

Today Rosa invented Flickritti (Flickr graffiti) and more specific, Flickr invaders. Whilst chatting with the Geeks 4 World Domination (g4wd) posse, she brought this up and my mind started wandering. Couldn’t I make a crawler that leaves notes on semi-random Flickr photo’s? Wouldn’t it be possible to create a trail with secret notes, maybe even constituting a work of art (whatever that may be) if seen in a larger picture (i.e. when the trail is reconstructed and visualized as one)?

Long time a go I wrote smile, a little script that fetches a random smile from Flickr. Although I had to study I figured I could easily tweak it to search for more words, e.g. those of a poem. So, without further ado, and without any trails and notes, I present to you: the Visual Haiku machine.

The Visual Haiku Machine gets semi-random images for each word in a haiku. It queries Flickr for a picture tagged with that specific word and picks one out of the 300 ‘most interesting’ pictures. These are then displayed underneath the word.

Alright, the haiku’s are over-romantic but I really like the fact that you can read a couple of lines over and over and get a different feeling with it each time, because the pictures always change.