Screen Printing with AI
I wanted to make a poster using artificial intelligence as a “tool”. I was thinking of different methods to make it. I started by using an image rendered in blender using photogrammetry from google maps, thus adding lighting and fog (volumetric fog) ; I also rendered some compositing data (normals maps, UV, mist) which will help me to rearange the image later.
Then, I passed the image in a deepdream filter (using tensorflow) and I also passed the image through an “face detector” using OpenCv’s Haar Cascade system. and I printed each coordinate of detected face into an text file.
With all these files I tried to combine, mask/show in order to get a nice poster :
I used a bichrome, dark blue and green with the idea of following a part of the Viridis Spectrum (colormapping used in MatplotLib to shows data).
Word2vec’s world