This Algorithm Can Turn Any Image Into a Van Gogh
kaishengtai/neuralart via GitHub
Thanks to an algorithm, any image can now look like it was painted by Vincent van Gogh—or any other painter, for that matter. Researchers from Tubingen, Germany have designed a program that can recreate the style of whatever painting you want it to, creating Cubist Picasso lookalikes or Starry Night-like impressions in just an hour.
In a yet-unpublished paper posted to arXiv.org, the researchers describe a system that uses object recognition to recreate the style of a particular image. The researchers were able to separate the painting’s visual content with the overall style of the painter, allowing the algorithm to generate similar aesthetic qualities in an image without reproducing the actual content of a painting like Starry Night. Instead, you can create a portrait of the Eiffel Tower that has been van-Gogh-ified, or a landscape in the style of Munch’s The Scream.
When this will be turned into a web app so we can all make our vacation photos look like custom portraits by a 19th century artist, we cannot say. For the computer science savvy, the researcher’s go into a little more depth about their programming on GitHub.
[h/t: The Guardian]