A new website using artificial intelligence (AI) technology and sophisticated algorithms enables users to turn their faces into images in the style of Leonardo da Vinci’s paintings. The Da Vinci Face platform invites participants to send in self-portraits which are then transformed into “Leonardo-style” images.
The virtual platform was created by Mathema, a small-scale tech business based in Florence. “We decided to revive the splendour of the Italian Renaissance, recreating portraits in the style of the Grand Masters of the past through the most recent and sophisticated AI technologies. We decided to start with most famous polymath genius in history: Leonardo DaVinci,” says Massimiliano Bellini, a spokesman for Mathema.
He explains why the project was challenging. “Leonardo's style is particularly hard to capture due to the widespread use of chiaroscuro, a pictorial technique where shapes are formed directly from colour instead of having a prior drawing that is, then, coloured.”
The company subsequently carried out extensive research on Da Vinci and on AI techniques. “The former was relatively easy for us, having at hand in Florence plenty of sources from curators, books and articles to academic investigations. The latter was not so easy due the fact that Leonardo’s style could not be captured with ordinary Neural Style Transfer techniques.” Mathema says it has pioneered a new “deep learning technology” using 500 million parameters linked to machine learning algorithms.
The Da Vinci Face project is self-financed by Mathema, Bellini adds. “Subsequently we received funding from some cultural associations in Vinci, the town in Tuscany where Leonardo Da Vinci was born. We are currently looking for further funding to extend the project.”