Being able to tell if a video is fake is about to get a lot harder.
From Snoop Dogg and Mike Tyson having a conversation dressed in drag,
— WHOOARMY.COM (@DJWhooKid) June 1, 2019
to Sylvester Stallone’s face plastered over Schwarzenegger in Terminator 2,
— Mikael Thalen (@MikaelThalen) June 3, 2019
we’ve seen deep fake edits are becoming creepily realistic.
Now, researchers at Stanford are revealing a new algorithm that allows video editors to modify footage of faces speaking – by deleting, changing or adding words that were never there.
The results are highly convincing re-edits that appear nearly undetectable from original or source footage.
Scientists say they hope the technology will help video directors and editors easily fix lines flubbed by actors and others on-camera.
“Visually, it’s seamless. There’s no need to rerecord anything,” said Stanford postdoctoral research scholar Ohad Fried.
Stanford University News described the process:
Should an actor or performer flub a word or misspeak, the editor can simply edit the transcript and the application will assemble the right word from various words or portions of words spoken elsewhere in the video. It’s the equivalent of rewriting with video, much like a writer retypes a misspelled or unfit word. The algorithm does require at least 40 minutes of original video as input, however, so it won’t yet work with just any video sequence.
While Fried acknowledged the potential for malicious actors to use the technology to deceive the public, he feels the need to have the technology is greater.
“Unfortunately, technologies like this will always attract bad actors,” he said. “But the struggle is worth it given the many creative video editing and content creation applications this enables.”