It’s been almost a year since Prism flew the head of all, with your app can turn any photograph into a portrait of almost any artistic school. Now, a new project at the University of Berkeley shows that this was only the beginning of a new era of platforms with great potential.
The research project just published, under the title of Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks, has been developed by engineers of the University of Berkeley, and shows the performance of this new platform, able to use a technique known as image style transfer Impressionist paintings into portraits that look real.
This “reverse Prism”, used as Medullary core his system an artificial intelligence, trained from the comparison of thousands of Flickr photos and paintings, in a system of contrast in the study as adversarial networks.
This network in the end managed to delineate the General bases of these filters, to determine dimensions, shades, depths and lighting that achieve the realistic effect of these canvases turned into realistic images.
An interesting factor is that in addition to all this the Berkeley system also can do the same as Prism and even make some images to alter elements within the portraits.
It’s a project that could represent the future of this kind of mobile applications.
But it will be some time before it reaches our smartphones.
Follow us and comment on Facebook.