Unglitching the images…

It is not a surprising aspect of “the glitch art” to be subject to endless debates of definition, as in its nature lies a paradox of indeterministic manipulation of a possibly deterministic reality. The debates circle around either the “glitch” aspect of it, or the “art” qualification, however an interesting debate may also rise from the “the” in this title. Is there a common feature among the grand multiplicity of works, styles, and methods that can collect all of these products into a category (adding to that maybe even the photographs of broken smartphones or crashed billboards representing a similar sense of aesthetics)? There may indeed be a common feature in these works. They all exemplify “products” (video, image, sound), which come out of semi-controlled “processes” (procedural and/or accidental), transforming a “living input”. The input is alive, because it exists, and it has meaning, even if very little, and it has an inertia – a potential to remain in order through time. Destruction only appears as such when it deviates the reality away from the predicted.

(Ramiro Serna, 2015)

My eyes find the most exciting pieces of meaning in the process itself. You don’t always have to understand the process in order to enjoy it, but any understanding certainly multiplies that joy. In some compositions you find meaning in the output because you are able to reconstruct back what was the “input” and visualize it in connection to the actual scenery.

(Ramiro Serna, 2015 / YY, 2015)

I am most impressed by the spirit of collaboration and iterative work in the glitch communities. This is not a coincidence, the nature of the destructive arts as described above can allow the works of the artists stack like dominoes, moving in very different, sometimes unimagined, directions. Nothing, and nobody can constrain the direction that the iterative process moves (or branches), and one of those directions may as well be towards the origin, re-approaching the orderly input.

(U. E. Deniz, 2014 / YY, 2015)

This inspired me to attempt to reconstruct some simple glitched images, the ones where the effects are dominated by horizontal shifts of slits. This is easy (possible) to somehow reconstruct, as long as one relies on the “continuity” of the vertical structures. The code referenced here basically picks each horizontal slit in the image and shifts it until it achieves the least color difference with the slit(s) above. Based on how far the reference slit is, one can apply weights in the sum so that the closest ones dominate.

(Stephen Irving, 2015 / YY, 2015)

For some images, it’s better to switch to grayscale, so that sharp color changes don’t affect the difference calculation severely. Of course, it has very little guarantee to actually work in any image, and sometimes it’s particularly fun that it gets a passing grade without the top score.

(Stephen Irving, 2015 / YY, 2015)

Once you find a new toy like that, you want to play with it in all possible ways, with the most challenging images…

(Ramiro Serna, 2015 / YY, 2015)

… or pretty ones

(Ramiro Serna, 2015 / YY, 2015)

Also interesting is when this horizontal shift is only one aspect of several different types of distortions on the image, and one approaches not the origin but a station in the middle.

(Selime Goc, 2014 / YY, 2015)

And of course, trying to reconstruct something which is not so badly broken may yield tragic consequences…

(Alkarex Malin via Wikimedia Commons / YY, 2015)

… or very little effect…

(E. G. Martinez, 2012 / YY, 2015)

After having some fun with this algorithm, it is good to talk about how we have manipulated the output result by calling it a reconstructed image, while what we were doing is only about straightening it. The prior assumption of the straightness that defines the method of reconstruction here is actually a good illustration of social engineering through feedback cycles of glitch and reconstruction in the flow of information in large scales. I believe this discussion is one of the things that increase the value of works in glitch arts, and it is sure that it will be recurring with many other creative approaches.

(A note on the code) Notice that it would be very easy to insert one more x-shift here to centralize the resultant image, however I preferred to leave that to anybody who wants to add a couple of more lines into the code. The Processing2.0 script is here:

https://github.com/yetkinyilmaz/processing/blob/master/Rekonstrakt/Rekonstrakt.pde

Enjoy…