Watch: New face tracking software enables mirroring expressions onto any YouTube talking headAuthor : AZIndia News Desk
Mumbai, Mar 21(AZINS) A team of researchers from Stanford and the Max Planck Institute for Informatics and the University of Erlangen-Nuremberg has just put new meaning to the phrase “Putting words into somebody’s mouth.” Using a sophisticated facial recognition software that tracks a subject, they have succeeded in mirroring expressions on a conventional YouTube video, in real time.
The system works with an off-the-shelf webcam pointed at the subject, where the person’s facial expressions--grins, eyebrow raises, mouth movements et al--are ‘transferred’ in real time to the target in the YouTube video.
This is simultaneously fascinating and scary--with the ability to modify a video of a person as it happens, this technology could pretty much change the face of video documentation, no pun intended.
In this demo video, researchers perform a real-time re-enactment using a regular YouTube video of US President George W. Bush, Russian President Vladimir Putin and Republican presidential candidate Donald Trump. No matter what the subject does--from goofy eyes to nonsense mouth movements--these get perfectly transferred to the subject as it happens.
Unlike many video special effects that tend to look unrealistic and sometimes awkward, these appear spot on with no visible sign of any special effects modification present. Even nuances like a slight twitch of an eyebrow or subtle twitches are captured and re-mapped to the target.
Fun as this appears, the researchers have not made this particular application public but are reportedly thinking about commercializing it, given its apparent efficacy.
On similar lines, popular face-tracking app Masquerade was recently acquired by Facebook. With the processing power of smartphones only on the upswing, it seems face tracking is the next big thing in the consumer video space.