Connor McGaffin – Looking Outwards – 02

This week I looked outwards to the work of Memo Akten. This work explores generating visuals of nature through the movement of manmade objects. The visuals generated have a paint-like quality, possibly driven by Akten’s admiration of impressionist art. Among the images seen in this video are simulations of crashing waves, flames, clouds, and plant life.

Atken describes the project as:

 “a deep neural network making predictions on live camera input, trying to make sense of what it sees, in context of what it’s seen before.”

With this information, I would infer that the algorithm used in this project may be similar to those used in Photoshop’s filters and patch tool. The filters apply an aesthetic quality to the visuals while the patch tool interprets what visuals are around it in an effort to generate more of the image. From watching the video, it becomes evident that certain hues of color correlate to different textures and hues, depending on which “mode” the generative algorithm is in.

A deep neural network making predictions on live camera input, trying to make sense of what it sees, in context of what it’s seen before. It can see only what it already knows, just like us.

I feel incredibly drawn to this work because of it’s ability to make natural phenomena feel so understandable. The work draws parallels between the incredibly organic and the accessibly human. I am most excited about the experiment with the water simulation driven by manual fabric manipulation.

 

source 

uploaded April, 2018