Wombo is a popular lip sync app in Europe and the U.S. It has been downloaded 2 million times in just a few weeks, and when you put in a photo, the person in the photo can sing a song, but the only shortcoming is the limited song templates. At the root, Wombo pop video still can not be separated from the mouth synchronization, Deepfake technology, only Wombo let this technology fly into the ordinary people, so that more people can make their own ghostly dynamic photos through the app one click.
By developing the latest templates, programs like Wombo allow users to blend together favorite characters, popular songs, elaborate choreographed dances, public figures and more, so you can imagine that the future of Deepfake will inevitably take the path of templating.
You’ve probably seen a lot of Wombo videos on social media, and Deepfake’s world is never short of hilarious and beautiful productions, with inter-dimensional fusions that make the corners of your mouth turn up like crazy!
When I was a kid playing the fighting game “Street Fighter” Ryu sang “Witch Doctor” three leaders of the Federal Reserve sang Rick Astley’s 1987 European and American gold song “Never Going to Give You Up” brainwashing?
The exaggerated facial expressions combined with the ghostly animation effects look both silly and funny, fully demonstrating the current state of Deepfake technology, which we can also call “earthy to the extreme is hip”. After playing it a couple of times, many people already want to make their own or their friends’ photos move. Open the door to lip sync brainwashing and quality output abounds.
Twitter users are using an app called Wombo to create lip-sync videos of their favorite characters. The app, available for free on the Apple App Store and Google Play, uses artificial intelligence to animate and lip-sync still photos. wombo’s official website initially featured only realistic photos of real people lip-syncing, but it didn’t take long for users to come up with other ideas.
Now, Twitter is flooded with videos of avatars faking popular songs such as Doja Cat’s “Boss Bitch”, Gunther’s “Ding Dong Song” and “Numa Numa”, and Wombo is easy to use. Choose a positive image of a character to upload, and once you’re done, all you have to do is choose a Meme song, and the image will start lip-synching to the rhythm of the song.
You can see that these videos are not from the game developers. The videos are usually pretty pixelated, but with a little work they can look pretty good, with the characters shaking their heads and raising their eyebrows more dramatically than you’d expect.
Come see how AI, Meme and Deepfake rub off on each other
Wombo’s name comes from eSports slang, the popular Nintendo hit “Super Smash Bros.” to be exact. If a player lands in a frenzy of action, someone will yell “Wombo Combo! Wombo Combo!”
The founders did some exploratory research into the origins of the slang term Wombo and discovered that it was coined by pizzerias who would put a ton of heavyweight toppings on their pizzas and call it the Wombo Combo.
After a short development process, Wombo launched in Canada late last month. “I had the idea to create Wombo in August 2020 while smoking on the roof with my roommate,” said Zion Bensin, Wombo’s creator and CEO, who called the release of the product itself a tremendous amount of fun. “I’ve been looking at the AI space, looking at meme emojis, looking at Deepfake, and finally we saw the opportunity for Wombo.”
In just a few weeks, Wombo has been downloaded around 2 million times. Players have also used Wombo to animate characters in games such as League of Legends, Fallout:New Vegas and Dragon Age.
The founder talked about how Wombo really works by using predefined dance moves to morph faces. He and his team shoot base videos for each song in his studio, and then use those to make the pictures move. He says:Â “It’s really about shifting, taking the moves from the face of the base video and applying them to your photo.”
Wombo can work its magic on anything that looks like a face, and it can create surprises on many things that don’t look like faces. It can work with all kinds of character images, with 3D characters working better; some people have tried some 2D animated characters and the videos just don’t look as smooth as 3D video game characters.
Currently, Wombo only offers 14 short clips of songs sung in counterpoint, but their recent success has them planning to expand their song library soon.
The importance of template innovation comes to the fore
Wombo’s emergence is also a prime example of the fast-paced world of AI research, where new technologies can become consumer products in a matter of weeks.
Like TikTok, Wombo offers a service that appears to help ease authors’ concerns about copyright infringement. Already, Bensin said, artists are actively wanting to put their music on Wombo, which amounts to an additional revenue stream for artists on top of the paid-first and de-advertising services.
“It’s a whole new way to attract traffic and a marketing tool for future pop song bookings.”
Artificial intelligence forgeries are no doubt becoming faster and easier, but the more convincing they are, the more work needs to be done. For example, the fire Tom Cruise’s Deepfake video mentioned last week (click to read) required not only an experienced visual effects artist, a top-notch stunt double but also weeks of preparation.
In contrast, the Deepfake produced by zero-based players still looks clearly like the output of an app like Wombo, and in the short term, Deepfake’s product is still only laugh bait. On the flip side, Wombo isn’t the first app to use machine learning to create a quick and funny Deepfake; other companies actively exploring the market include ReFace and FaceApp.
As the latest example of a growing trend, it highlights more the rise of Deepfake app templates that allow users to blend together favorite characters, popular songs, elaborate choreographed dances, public figures and more.
The result is that the future of Deepfake is bound to be templated. The key to the Deepfake market is not only to make it “accessible”, but also to make the templates and songs new and interesting.