Untitled (2020). Taking procedural morphogenetic modeling a step further, we have trained a GAN with a series of algorithmically created 3d models. A human actor thus effectively designs visual seeds and constraints of potential. The machines take it from there, breeding new forms that take us a step further into the world of dreams and possibility. The hybrid results go beyond networks trained on real images or their painted depictions, since the training set is already a step removed from our lived reality.