Samsung researchers have developed a new software that requires just a single source image to create “living portraits” using just one photo of a person’s face.
Photo Credit: Egor Zakharov
The program works by using a dataset of multiple talking head videos featuring people with a wide variety of facial features. Researchers then trained the program to identify the “landmark” features of those faces: the eyes, the shape of the mouth, and their noses. Instead of pasting one face onto another and using the expressions of a specific person (which is what most deepfakes currently do), the new program uses common human facial expressions to puppeteer a new face on its own. It even works on the Mona Lisa!
Hopefully, they don’t figure out how to make the hair look real… or else.
The post This Samsung AI Can Create a Fake Video of You From Just One Photo appeared first on UberFacts.