TikTok-linked AI project creates deepfake video from single image

Deepfake it ’til you make it.

In the very near future, a single snapshot from your phone and an audio track plucked from the internet will be all you need to create a shockingly real AI video that can fool viewers into thinking you’re as smart as Albert Einstein.

That’s according to a newly-released preview of the jaw-dropping new AI model said to be forthcoming from ByteDance, owners of TikTok, which revived the deceased E = mc² declarer to demonstrate the futuristic tool’s cunning capability.

OmniHuman promises to allow everyday people to sing like pop stars, average kids to sound as smart as philosophers and much more, its creators promised.

“OmniHuman significantly outperforms existing methods, generating extremely realistic human videos based on weak signal inputs, especially audio,” an online introduction to the project stated.

“It supports image inputs of any aspect ratio, whether they are portraits, half-body or full-body images, delivering more lifelike and high-quality results across various scenarios.”

The creators showed their work earlier this week on online open-access archive arXiv, Forbes reported.

While the public doesn’t yet have access, multiple sample videos give a taste of what’s to come, showing the subjects from all angles — and making all kinds of body and hand movements that in most cases appear shockingly true to life.

The astonishing video of Einstein shows the German-born genius gesturing enthusiastically in front of a chalkboard while giving a speech about the value of art and emotion in art.

And while the quality of the visuals are top notch, the words are not his own. The audio comes from a speech given by neuroscientist and animal emotion researcher Jaak Pankseep over a decade ago at a TedXRainier event in Seattle.

“They’re very impressive,” Freddy Tran Nager, a clinical associate professor of communications at the University of Southern California’s Annenberg School for Communication and Journalism, said of the clips in an interview with Forbes.

“If you were thinking of reviving Humphrey Bogart and casting him in a film, I’m not sure how that would look. But on a small screen, especially on a phone, these are impressive,” he said — floating the idea that in the near future, students could choose, for example, to have Marilyn Monroe teach them statistics.

The tool puts ByteDance very much in the running as the race for realistic AI heats up, Forbes suggested — with expert Nager even floating the idea that the company might be looking for a way to replace influencers.

“TikTok can say, ‘You know what? Now we can just create videos on our own. Who needs the human beings?’,” he said.