Visual deepfakes – “I’ll believe it when I see it”?

Deepfake is a technology capable of altering or creating from scratch what is visible on screen. It can be divided into several levels, ranging from simple facial retouching in photographs to the creation of entirely new people who speak and move in a video as if it were a real recording.

Among the various variants of deepfake visual technology, we can distinguish:

• Identity swapping – here, the computer superimposes one person’s face onto another’s body, ensuring that it matches the lighting and perspective. This technique can be used in the film industry, for example when filming scenes involving stand-ins.
• Facial reconstruction – this enables the precise recreation of the facial expressions, eye movements and lip movements of the person visible in the footage. It is also sometimes useful in filmmaking (e.g. when dubbing voices for animated characters). However, fraudsters are also fond of it: they can deliberately add wrinkles or dark circles under the eyes to a person in a recording, or alter their facial expressions to make them appear extremely exhausted, ill or tearful.
• Altering individual facial features and characteristics, such as eye colour, nose shape, wrinkles, and even combining the faces of several people into one.
  Creating a completely new character that does not exist in reality.