Vincent van Gogh's artistic genius took the art scene by storm shortly after his death. The tortured artist's unique vision of the world still captivates today.
Now scientists from the Bethge Lab in Germany have demystified how the influential painter interpreted the world using an artificially intelligent (AI) system that can learn any artist's style, like the swirls and dots that are characteristic of van Gogh's work, and replicate it on other images.
According to Leon Gatys, PhD student and the lead on the paper published in the open-source journal arxiv, this is the first "artificial neural system that achieves a separation of image content from style."
To do this, the AI system had to first be "taught" the features of the famous painting before replicating it. The scientists had it analyze van Gogh's most famous painting, "Starry Night."
The scientists fed the painting into the system, which is composed of stacked layers of computing units that imitates the interconnected structure of the cells of the brain. The program analyzed the different layers of color and structure in the painting to discover van Gogh's painting style.
They also had to train the AI with the image it was supposed to paint in van Gogh's style. They used a photo of a river in their hometown, the Neckar river in Tuebingen, Germany.
The AI works like an assembly line — each layer is responsible for one thing. The lower layers identify the painting's simple details like dots and strokes. Upper layers recognize more sophisticated features like the use of color.
The figure below shows how each layer of the AI took the style of the van Gogh's painting detail by detail and applied it to the photo of the river.
The result looks as though van Gogh was standing at the riverbank in Germany rather than at his window in Saint Remy de Provence, the original setting for "Starry Night."
The AI didn't just take colors and place them in corresponding areas on the photo. The program was able to make sense of the original photo's shadows and highlights and actually 'understand' the scene, in a sense.
Gatys wrote that the AI gives us a mathematical basis for understanding how humans, including one of the most influential artists in the world, perceive and create art because the program can mimic biological vision and the human brain.
What van Gogh actually saw or thought while painting his works of art is still elusive, but the algorithm gives an interesting view into the patterns that influence his creations.
Join the conversation about this story »
NOW WATCH: This breakfast burrito may actually help you lose weight