Engineers from Microsoft Research Asia have created an AI technology capable of turning a single still image of a human face into a fully animated (and editable) talking head in real-time. Am I going to turn my senior yearbook photo into a news anchor? As long as the technology works with grainy black-and-white images!
Once the still image is scanned and an audio clip is added, the face will automatically sync its mouth and facial movements to the words being spoken. Users can then tweak the resulting video with sliders to change the angle of the head and distance to camera, gaze direction, and even emotion (e.g. happy, angry, surprised). Deepfakes just got a whole lot deeper.
With the ability to make a convincing video of anybody saying anything in real-time from nothing more than a photo, soon we won’t be able to tell whether what we’re watching is real. Or has that time already passed? I feel like it may have already passed, at least based on all the videos my relatives send me thinking they’re real. It’s like not a single one of my aunts or uncles has heard of CGI before.
[via TheAwesomer]
With over 17 years of writing experience, Jonathan has had a passion for all things tech-related, ever since watching Inspector Gadget as a child. He's disassembled countless appliances and managed to put a few back together, and one still works. When not writing, he can often be found playing video games or accidentally hurting himself in the garage.