Algorithm Transfers Famous Artists' Styles to Scenes from Movies
A group of researchers at Germany’s University of Freiburg have created a computer algorithm that can mimic the style of any visual artist and transpose it onto a movie scene. The algorithm, which uses a deep neural network to analyze the distinct style of different artists, is an impressive piece of technology—and the movie-painting hybrids it produces are strangely beautiful.
MIT Technology Review explains that researchers faced two main challenges: First, developing an algorithm that could distinguish style (brush strokes, color schemes, shapes, and so on) from content, and second, mapping those stylistic elements onto a moving image. Researchers discovered that they could capture an artist’s style by looking at connections between different artistic elements. “The way an artist reproduces a face is correlated with the way he or she reproduces a tree or a house or the sun,” explains MIT Technology Review. “Capturing this correlation also captures the style.” The researchers published their findings in a paper called “Artistic Style Transfers for Videos.”
Watch above as famous movies and TV shows are blended with the works of Vincent van Gogh, Henri Matisse, Pablo Picasso, and more.
[h/t MIT Technology Review]
Banner Image Credit: Computer Vision Freiburg, YouTube