Taking a Query="line" for a Walk (2008)

Taking a Query="line" for a Walk is a software-based animation that renders selected images from Flickr.com which are tagged with the word "line" as a never ending moving "line of lines." A single "line" within each image is determined manually and saved to a database, which custom software uses to align and animate each of the images according to its corresponding determined line. Other than rotation and placement, the images are not manipulated in any way. The animation is projected on a corner to emphasize the illusion of a continuous line.

Video

Context

Paraphrasing Paul Klee, the piece engages the idea of "line" both formally and in the realm of language and the construction of meaning. The notion of a "tag" is interesting to me because of its "limited" bandwidth, yet numerous meanings; if we tag something with "line", what do we mean? In some cases the meaning is simply formal (such as in the multiple architectural images that are in the database), while in others the word may be used to describe its subject (clothes line, power line, subway line). Tagging defines, but it also shows the limitations of a language that tries to define atomically.

In addition, this piece is part of a series of explorations into what it means to create artworks out of search engine queries. The piece adheres strictly to results found only through the given query: no "navigation" has been performed in order to obtain the images, only selection within the presented results. If search engines are the new arbiters of knowledge (what cannot be found will rarely enter into the public consciousness), then they are also the arbiters of the meaning of the words used as queries, even though such search engines cannot really discern "meaning" as understood by humans. Perhaps the idea of "relevance" is the computer equivalent to human "meaning."

Development

Custom tools were developed in Java to gather, select and manipulate the images from Flickr (These tools have since been ported to Cocoa). The "line" in the image was defined by "drawing" an invisible line over each picture, whose start and end points was recorded to an XML database, together with the physical location of the image file. Another tool was developed to permit a rapid and visual arrangement of the images in the database. The arrangement of the images is not random: they have been grouped thematically, with attention to formal qualities to provide the illusion of a smoothly flowing line. For performance reasons, the final display is handled by an OpenGL application written in C++ with the JUCE toolkit. For the truly curious, the tools to gather and manipulate the images took the lion's share of the development time of the piece, while the final display software was a relatively simple ordeal. The only gotcha in the display system is that since the database is rather large, not all of the images can be loaded into memory at once; hence, a mechanism to dynamically load the images before they were to be displayed was required. To permit a smoothly running animation, this mechanism runs on a separate thread of execution that attempts to keep a buffer full, allowing the software to run without many "hiccups."