With this new development from Nvidia Research , it is possible to turn a simple artisan drawing into a real photo thanks to the artificial intelligence it uses.
The Nvidia Research algorithm uses artificial intelligence to develop images that emerge from a simple sketch.
An interactive app called "GauGAN" uses antagonistic gene networks (known as GAN), to convert simple scribbles into realistic images. The name was chosen to honor the post-impressionist painter Paul Gauguin.
This Windows-style "Paint" software allows you to even manipulate a scene by choosing names of things instead of choosing colored brushes.
If, for example, a landscape with trees is drawn, it is possible to change the scene simply by modifying the label "boil" to "snow" to create a winter image. This development also admits to put filters of styles of different artists or even of different hours of the day.
Bryan Catanzaro, vice president of research at Nvidia, said that it is very easy to brainstorm designs with simple sketches or scribbles and then with "GauGAN" turn them into photorealistic images.
At the moment the development is not finished, but this technology was tested with thousands of photos and each color was related to an element of nature. "GauGAN" was designed for researchers, rather than for general users and can be used in AI Playground and with other apps when it is finished.
What problem can be presented in the future? ... that this "democratization of art", leads to have similar images always.