Video evidence is about to become a lot less reliable.
Computer scientists at Stanford University, the Max Planck Institute, and the University of Erlangen-Nuremberg in Germany have created software that can create a 3D model of a face from just 15 seconds of video. Then, with a webcam, they can make that face do — and, most incredibly, say — just about anything.
The 3D nature of the computer-generated faces is reminiscent of Snapchat's new face swap feature, except it substitutes facial movements rather than features.
In the video's YouTube description, the team notes that this technology has existed in feature films for 30 years. However, their software makes it possible to achieve similar computer-generated videos with just a consumer-grade computer and pre-recorded videos.
Several commenters have raised the frightening prospect of what would happen if this technology got into the wrong hands. But the research team emphasizes that they are working to make it apparent if a video has been edited, so that inauthentic clips can be easily identified.
"Our primary goal is to create a mathematical model of our world," researcher Matthias Niessner told A Plus. "Such a model enables computers to reconstruct, understand, and interact with it."
That being said, this is an important reminder of how high-end video software could potentially lead to a proliferation of misleading videos. As this technology continues to evolve, we will all need to be that much more vigilant about verifying and sourcing the content we come across online.