ΌAs the years go by, artificial intelligence will evolve along with its capabilities, since Nvidia with the use of AI has managed to create faces of people who are not real but are the result of mixing many different, combining their characteristics.
The video illustrates the results of this mix of features from human photos and shows how much AI has evolved so that it can "learn" and create photos of people who are not real, but a product of the abilities of artificial intelligence, and not we know where they will go in the future.
The collage of photos shown in the video was created by artificial intelligence and more specifically by a team of Nvidia researchers, who used a type of AI called Generative Adversarial Network (GAN) with impressive and very realistic results.
The title of the research is "A Style-Based Generator Architecture for Generative Adversarial Networks" and researchers Tero Karras, Samuli Laine and Timo Aila from Nvidia, show that they redesigned the GAN architecture with the "style-based design" method. According to it, AI automatically learns to distinguish the characteristics of people, from their posture to their hair and freckles.
After "training", the network begins to create images that do not seem to have come from the real photos used as a source, creating faces that are essentially young people who are the creation of artificial intelligence. The researchers then report that the network considers that each photo it creates is a collection of different "styles", with each affecting the creation of artificial intelligence photos, with researchers being able to "play" with each feature and to create a new face each time.
[the_ad_group id = ”966 ″]