- Redditor wants to know if he’s the a**hole for ghosting pregnant partner Thursday 8:19 PM
- How to go live on TikTok Thursday 8:08 PM
- Joey Salads suggests Democrats carried out Santa Clarita mass shooting Thursday 7:31 PM
- How influencers use TikTok to make money and launch careers Thursday 7:18 PM
- How to stream Argentina vs. Brazil live Thursday 6:51 PM
- How to watch Disney+ on a smart TV Thursday 6:28 PM
- Miss Fame calls out Justin Bieber for low music video appearance pay offer Thursday 6:19 PM
- Trump Jr. ranked No. 1 on best-seller list—after the GOP gave away copies of his book Thursday 5:45 PM
- How to get Disney+ bundle if you already subscribe to Hulu and/or ESPN+ Thursday 5:19 PM
- Mo’Nique suing Netflix for race and gender discrimination Thursday 5:09 PM
- Students outraged that professors accused of sexual misconduct are still teaching Thursday 5:00 PM
- TikTok users jokingly wear big hats to sneak snacks into movie theaters Thursday 3:59 PM
- Why today’s new facially recognition bill is being called ‘woefully’ inadequate Thursday 3:15 PM
- Facebook has given more user data to the government than ever before Thursday 2:57 PM
- How to sign up for Disney Plus Thursday 2:55 PM
A team of researchers with Samsung have found a new way of creating deepfake-style videos based entirely off of a single image.
In a paper published online this week, the team, hailing from Samsung’s AI Center in Moscow, explained how unlike most methods, which attempt to digitally paste one face onto another, their fake videos are created by teaching a computer to understand universally common facial movements.
Referred to as “few-shot learning,” the new algorithm was developed after initially feeding a program thousands of celebrity images and videos in order for it to learn about the human face. Once equipped with such knowledge, the program can then apply what it has learned to a single image it has never seen prior.
The method allows for increasingly realistic videos depending on the number of images. While traditional deepfakes require anywhere from hundreds to thousands of images of a specific individual, the team’s algorithm can use anywhere from one to 32 for comparable results.
A video published on YouTube Tuesday shows the new algorithm in action.
While far from perfect, the video reveals how the technique can be used not only for living people with a wealth of source imagery but even for individuals whose likeness is not as well documented.
In one such example, the team is able to animate the face featured in the Mona Lisa painting. A singular photograph of iconic American actress Marilyn Monroe is also brought to life.
The method represents yet another step forward for fake video creation, which has become increasingly popular and accessible in just the span of a few years.
- How to spot a deepfake
- Elon Musk is a giggling baby in the latest terrifying deepfake
- This deepfake takes Bill Hader’s Schwarzenegger impression to the next level
Mikael Thalen is a tech and security reporter based in Seattle, covering social media, data breaches, hackers, and more.