Article image Samsung brings Mona Lisa 'to life' with deepfake AI
May 27 2019
The researchers note that just using one picture makes the software less effective, you can see the animated versions all have distinct "personalities" they're deriving from the people they're being modelled onto. As the team proves, its model even works on the Mona Lisa, and other single-photo still portraits.
The results are indeed mesmerizing, as you'll see in the video at the end of this post.
"Such ability has practical applications for telepresence, including videoconferencing and multi-player games, as well as the special effects industry", write the researchers in their paper. As researchers continue to come up with low-lift methods for making high-quality fakes, there's a concern that they'll be used against people in the form of propaganda - or to depict people in situations they'd object to, like pornographic videos.
Samsung's algorithms were trained on a public database of 7,000 images of celebrities gathered from YouTube. Enter Samsung's new research, in which a neural network can turn a still image into a disturbingly convincing video.
The system makes use of a convolution neural network, a type of neural network based on biological processes in the animal visual cortex.
The number of images taken can also add to the realism of the movement. It's easier pulling the trick off with actual images of a person rather than paintings, as photos are a lot closer to reality, but the Mona Lisa deepfakes are still quite impressive. The video had some weird artifacts which should be looked at. GAN is a technique that learns to generate new data with the same statistics as the training set.
On the plus side, your favourite movie and TV stars never have to grow old and die - AI similar to this is soon going to be smart enough to produce fully realistic performances from just a few photographs, and in record time, too.