The Next Rembrandt is an original portrait created with machine learning algorithms trained using the Dutch master’s works. The resulting image isn’t a plain old bitmap, either, but a fully three-dimensional artifact built with scans of real paintings’ brushstrokes and protrusions.
“We really wanted to understand what makes a face look like a Rembrandt,” Emmanuel Flores, director of technology for the project, told the BBC.
After they had been digitally tagged by humans, data on Rembrandt’s paintings was gathered by computers which discovered patterns in how the Dutch master would, for example, characteristically shape a subject’s eyes in his portraits.
Then, machine-learning algorithms were developed which could output a new portrait mirroring Rembrandt’s style.
To limit the many possible results to a specific type of individual, the computer was asked to produce a portrait of a Caucasian male between the ages of 30 and 40, with facial hair, wearing black clothes with a white collar and a hat, facing to the right.
The involvement of human artists in the final work is unequivocally denied: “humans didn’t decide the final look and feel of the final portrait – they simply chose algorithms based on their efficiency and let the computer come up with the finished result.”
The suggestion of emergent brilliance from the machine, then, is quite exciting. That said, there are an awful lot of Rembrandt portraits in exactly this strictly-composed style – a good place to get started.
