Samsung’s deepfake AI fabricates a motion clip of from a single photo

0
407
Samsung’s deepfake AI fabricates a motion clip of from a single photo

Believe it or not, but the new Samsung technology has even turned the Mona Lisa into a living video. But don’t worry just yet, because the tech hasn’t reached the hands of any devious mastermind.

Samsung developed a highly capable deepfake technology

Software for creating deepfakes – fabricated clips that make people appear to do or say things they never did – usually requires big data sets of images in order to create a realistic forgery

But Samsung has figured out a way to use an artificial intelligence that can generate a fake clip by feeding it a little as one photo.

The primary objective of this new technology is for it to be fun. And, truth be told, seeing photos and even classic portraits come to life is pretty entertaining.

This new technology was developed by a Samsung artificial intelligence lab in Russia and they detailed it in a paper earlier this week.

The downside of this new AI invention is that these are the types of techniques that, when they rapidly develop, they can also create risks of misinformation, election tampering and fraud.

“Following the trend of the past year, this and related techniques require less and less data and are generating more and more sophisticated and compelling content,”

Farid said.

Even though Samsung’s process can create visual glitches, “these results are another step in the evolution of techniques … leading to the creation of multimedia content that will eventually be indistinguishable from the real thing.” according to Hany Farid, a Dartmouth researcher who specializes in media forensics to root out deepfakes.

How does the Samsung deepfake software work?

The new software works sort of like Photoshop only several times better… and a bit differently. It produces forgeries by using machine learning to fabricare moving, speaking human beings that are convincing enough.

Although the technology is extremely entertaining, making all sorts of jokes possible like the one that put Nicolas Cage‘s face into movies and TV shows he wasn’t in, it can also be insidious, such as when it’s used to graft an unsuspecting person’s face into explicit adult movies. As you probably guessed, it’s usually used in revenge porn.

Deepfake videos usually require a big data set of images to fabricate a fake video of someone, but an artificial intelligence system developed by Samsung created a fake clip from a single picture.

The talking heads genre currently used by Samsung

Samsung dubs its new creation “realistic neural talking heads”. The word “neural” is a nod to neural networks, a type of machine learning that mimics the human brain.

The researchers saw their breakthrough being used in a host of applications, including video games, film and TV.

The paper was accompanied by a video showing off the team’s creations, which also happened to be scored with a disconcertingly chill-vibes soundtrack.

The Samsung deepfake system doesn’t need lots of images to work

The Samsung system uses a trick that seems inspired by Alexander Graham Bell’s famous quote, “Before anything else, preparation is the key to success.” The system starts with a lengthy “meta-learning stage” in which it watches lots of videos to learn how human faces move. It then applies what it’s learned to a single still or a small handful of pics to produce a reasonably realistic video clip.

The issue is that the technology isn’t perfected to render the most accurate results, but this kind of system can quickly adapt to a new target person using only a few images without extensive retraining.