GauGAN neural network will transform your bad drawings into masterpieces

This isn’t the first time scientists are exploring artificial intelligence for improving photographs or creating portraits and still life paintings, but the GauGAN neural network is much more advanced than the others. This development offers startling possibilities.

GauGAN is a project presented by NVIDIA Research at the GTC 2019 conference in California. This neural network transforms sketches made in Paint beyond all recognition, making them incredibly spectacular and realistic. What sets this AI apart from similar programs is its attention to detail: paintings include even the smallest shadows and reflections from objects, and the neural network can also change seasons.

The project was named after French artist Paul Gauguin. The program was trained using millions of paintings as inputs. The artificial intelligence is made up of two subprograms: one creates the drawing, the other searches for inconsistencies and then corrects them.

There is still no news regarding when the neural network will be made publicly available, but computer game developers are already interested — they believe that the development could help them with location modeling.

https://www.youtube.com/watch?time_continue=104&v=p5U4NgVGAwg