Algorithm for Quickly Creating Deepfake Developed in South Korea
Developers from South Korea have created a unique algorithm with which anyone can create a high-quality deepfake. To do this, you need only one photo or a short video recorded on the front camera. That is, in other words, now it has become much easier to manipulate the images and videos of famous people. Important to note is that critics of this technology just talked about such a development of events in the past. Earlier learning the algorithm required more initial data: at least a few minutes of a video of a public speech filmed from different angles. But MarioNETte overtakes its competitors - just give it one photo or a short video of poor quality and the deepfake is ready. Besides, the quality of fake videos has become much higher and realistic. The algorithm consists of three parts: image recognizer, binding, and transformation algorithm. It is important to understand that technology develops every year and soon it will be possible to distinguish a deepfake from a real video only with the help of software.