Menikvadrati
The Menikvadrati series consists of drawings that I created during the Covid-19 lockdown, inspired by my experiences with frequent zoom-calls.
In the second round of Menikvadrati, I combined my drawings with neural networks to create hyper-realistic images of human faces. It was an attempt to use the power of AI to see how humans would look like if we assume that my drawings were hyper-real.
Each drawing was manually labeled using face parsing labeling templates, which are commonly used in machine learning tasks. This process involved identifying and labeling specific facial features in the drawing, such as eyes, nose, mouth, and hair, which could be later used by the neural network to generate hyper-realistic faces.
After labeling, the images were processed using a pix2pix neural network, which is a type of generative adversarial network (GAN) commonly used for image-to-image translation tasks. The network was trained on the pairs of labeled images and corresponding face photos to generate new faces based on the input drawings. The resulting images were a combination of the original drawing structure and the hyper-realistic features generated by the neural network. The Menikvadrati series thus explores the intersection of traditional drawing techniques and cutting-edge machine learning technology.
