Through the development of programmers from the University of California at Berkeley, the horses in the photos are transformed into zebras, and apples to oranges

it is too early to talk about the launch of a new app for smartphones but work in this direction is underway. By analogy with the notorious "Prism" (the app the usual stylized pictures under works by Picasso, Monet and other famous artists) American scientists have tried to register the reverse algorithm — turning these paintings into something similar to a real photo of the landscape. With the result of this transformation you can see in the picture above. The basis was taken landscape impressionist Claude Monet's "low Tide in Varangeville".

the Resulting photos are not an accurate reproduction of the state of nature in the moment when Monet wrote to his painting. Rather, they give a General idea of the landscape that served as the basis for paintings by Monet. Though not very clear why the artist inspired the kind you can't see in this photo of the coast at Varangeville.

the Next stage of researches of the American scientists is all the more mysterious in their purpose. They turned the photo from the painting, back into the picture. As if you did the translation from English into French, and then again translated the French text into English, hoping to get the text that is maximally similar with the original.

Next, artificial intelligence has been tasked to transform the image of an Apple in an orange horse — in Zebra, cat — in the dog, winter landscape Yosemite — in the summer, etc. the Result leaves much to be desired (see picture below). But some-как already working.

Why are all these experiments? Exercise scientists from the often seem like nonsense, but without them there is no progress. Similar problems on the conversion of impressionism in the photo allow you to test and improve artificial intelligence. In the future this will allow the machines to take care of the routine tasks on the attribution of works (for example, to narrow the circle of "suspects") and even to make rough conclusions about their authenticity. Yes, so far it looks fantastic. We periodically ask this question to our experts: how soon machine will be able to determine the authenticity on the analysis of technical and stylistic devices of the artist? Now experts-искусствоведы confidently answer that ever. For the solution of this complex task it is necessary to consider too many objective and subjective components. They say, "feel" the art close to the man machine is unlikely to learn. But Shazam 10 years ago was considered science fiction. And now any phone for a hundred dollars surely recognize "heard" music on the count of three. We often do not ask questions of knowledgeable people, and Google and Siri. Here you have Zebra-в-лошади, and oranges-в-яблоки.

by Maria Onuchina, AI

Sources: engadget.com, artinvestment.ru