Artificial Intelligence and deepfake have accustomed us to seeing beautiful ones: from Salvador Dalí welcoming visitors to a museum to the people in the videos who change sentences when we go to edit the text.
Recently an application of AI and deepfake hit us: is a Samsung team, and produces moving subjects from a simple photo. We talked about it here.
An evolution of this algorithm allows today to synchronize a piece of music with a photo made dynamic, producing impressive results.
In this video, for example, the well-known Russian mystic Rasputin, who died in 1916, has the posthumous pleasure of interpreting "Halo", a hit song sung by Beyoncé. Someone will laugh, others will be horrified. I have the ability to do both!
Here is the rather remarkable performance. As you watch it (it lasts a few seconds) ask yourself questions like me about the future of this technology. Because this technology, believe me, deserves all our questions.
Not bad the game, isn't it?
The researchers believe the model could be used to automatically create animated characters. It will also be able to "bridge" the time gaps during slow streaming, or simply "dub" the films by letting the characters speak directly in the chosen language.
For all brains here is the paper that illustrates the technique.