'Unknown Forms Of Face' looks for faces in random images.

A random image - usually looking like static noise - could possibly display anything, including a portrait of you. It just isn't very likely that you are going to see anything recognizable at all. However, when creating a large number of these images, it becomes more likely that some of them contain information distinguishable by the human eye - faces, for example. Running a face detection algorythm, those images are filtered out, and if you look (very) closely, you can see (human?) faces looking at you out of the gray randomness.

Unknown Forms of Face has been awarded a Honorary Mention at the Prix Ars Electronica 2015, category 'u19 - Create Your World'.

randomFace_1
Click into the image for more examples.

I find this a very neat metaphor for - well, life, after all. Being very improbable, but possible, it does become likely to happen if there are enough chances - so goes the theory. You just have to look in the right places. In case you don't like metaphors, or you like to view life differently, just see it as a program that finds faces in random images.

Things become very theoretical at this point, but consider taking this principle further:
Imagine a function-detecting algorithm, running over a huge set of randomly generated molecules, looking for drugs or materials with special properties. Even more interesting, imagine a life-detecting algorithm scanning an enormous database of randomly generated combinations of molecules! Like this, we can look for anything, knowing only some of it's properties, without the barrier of human presumptions (those familiar with the idea of tractability and the computational complexity theory will be less excited about this brute-force method, though. This has actually been an idea that sparked initial excitement in AI research, but failed to deliver useable results).

You can look for faces right here, in the browser. Just click into the frame below and start looking for the needle (a face) in the haystack (the mass of possible images in this frame)! Once a face is found, it will highlight it and stop the search.



Note that this browser-based application is relatively inefficient in looking for needles (pardon, faces), so you might consider checking out Unknown Forms Of Face's GitHub page for the desktop version!

https://github.com/DominikKoller/unknownFormsOfFace