The software engineers at Samsung have revealed how effective deepfake technology is becoming – by adding it to the Mona Lisa.
The famous Leonardo Da Vinci painting has been augmented to look like it’s moving and talking. And the results are pretty unnerving.
In a paper published by Samsung’s AI Center in Moscow along with the Skolkovo Institute of Science and Technology, researchers explain how they created the spooky effect.
They assigned a series of ‘facial landmarks’ to the portrait and then applied an algorithm that has access to metadata from a vast amount of image banks. They can then form different ‘movements’ of the Mona Lisa’s head.
They used paintings as the basis for their study to prove that you only need one good image to create an avatar using this tech.
It wasn’t just the Mona Lisa that got the deepfake treatment. The researchers also used their skills on portraits of Salvador Dali, Einstein and Vermeer’s Girl with a Pearl Earring.
‘We often accept things at face value, having neither the time nor the skill to check the authenticity or veracity of things ourselves. But the sophistication of Deep Fakes is such that even an expert could struggle to tell what’s real and what’s not,’ Futurist Richard Watson told Metro.co.uk as part of our Future of Everything series.
‘We’re not talking about ventriloquist dummies here. We’re taking about things that look, sound and feel utterly convincing. This might all sound trivial, and many of the examples are, but sometimes things can get more serious.’
Source: Read Full Article