GoogleSynth, new images from Google’s archives.

.art



07.02.03 GoogleSynth, new images from Google’s archives.


GoogleSynth is a software developed by Paul Andrews which generates a synthetic image using data from two other images taken from the Google Image Search engine. The author adapted a piece of code used for texture synthesis which uses the texture of an image and tries to reproduce another one with it, and automated its process of data acquisition to use images taken from the Net. Programmed for PC and MacOSX, the software is available for free download. The code is not very interactive, but the user can tweak a few parameters anyway. It selects two words from a list of 62.000 and uses them to find two images with a random choice algorithm. This work, as well as correlating the verbal ‘seeds’ with the visual results, produces its output through a generative reformulation of the material found on the Net. Even if it’s difficult to recognize the traces of the images it started from, the final result feels perfectly defined by the two chosen terms. Besides, the mutation of the images, mysteriously in harmony with the title, opens new possibilities for investigating the semiotic interaction between verbal relations and visual transformations.