3 ms·
It makes the use of stunt doubles easier as well since the actor's face can be mapped to the body double. And it isn't like this point in time wasn't predicted
by ChuckMcM 9y ago
It makes the use of stunt doubles easier as well since the actor's face can be mapped to the body double. And it isn't like this point in time wasn't predicted by just about everyone in CS. Given enough computer power and the right algorithm pretty much any data is fungible, video or otherwise.
But we've just entered the most dangerous period when most people still believe what they see in a video is what was originally videoed. It is when people will use this technology for deception and get away with it like they did with Photoshop before 'shopped!' came into the more general populations consciousness.
- dllthomas 9y agoI don't disagree, but I hope that the transition will be eased by the fact that the photoshop transition is within living memory of so many.
- HorizonXP 9y agoA cool, possible solution would be to have every participant in the video to digitally sign the final file to verify that they're actually in the video, consensually. That way, if anyone finds a deep fake video, it would be easy to dispute it since it doesn't have the person's signature. Of course, this would be exceedingly difficult in practice, since you'd need the signing to be opt-out instead of opt-in for it to be truly effective.