Robotart

Starry Night Reproduction 2 – Deep Learning Training – Van Gogh

One of last year's Robot Art judges commented, "A key point to keep in mind is that in every case the robot is a tool, more sophisticated than a brush, but depending fundamentally on the creativity and skill of the humans behind it.“ ----- CloudPainter's 2017 Robot Art entry aims to challenge this assertion by not only attempting to paint each canvas from start to finish, but also making as many meaningful creative decisions as possible along the way. Currently, the only decision made by a human is the decision to start a painting. CloudPainter then uses Deep Learning and Artificial Intelligence to do the rest from creating an original composition to deciding on the final brushstroke. While admittedly we make the ultimate decision of when an artwork is complete, CloudPainter is learning more with each new painting and getting better at making this decision on its own with every iteration. ----- CloudPainter is the father son team of Pindar, Hunter, and Dante Van Arman. For the past decade they have been experimenting with computational creativity by making painting robots. The specific details behind how their robots create original artwork using a variety of Artificial Intelligence, Deep Learning, and their open-source historic database with detailed brushstroke data for hundreds of paintings can be seen in the images below and accompanying video.