Steve Troughton-SmithHere’s your AI astonishment/nightmare fuel for today:
"TL;DR: single portrait photo + speech audio = hyper-realistic talking face video with precise lip-audio sync, lifelike facial behavior, and naturalistic head movements, generated in real time.”
🇸🇭🇮
@stroughtonsmith I get that this is worrying, but "hyper-realistic"? this woman is growing teeth in front of your eyes as she speaks 😅 (and that's not even getting into the quantum tangled hair that sticks in place as the head moves)
Jim Bliss@shi @stroughtonsmith Right now, yes. But as I said elsewhere, if you are viewing it low-res on a social media video on a small screen and only giving it half your attention then some people will fall for it. Especially people who don't realise it's AI and are not looking for telltale signs.
Then in 5 years it'll fool you, me and everyone else even on a high-res IMAX screen. This is a nascent technology that is already massively more convincing than it was 12 months ago.
@JimBliss I viewed the video on a low-res screen on social media 🤷♀️
also I think you'll notice I didn't comment on possible future developments, just calling the video uploaded above "hyper-realistic" - when (even at low-resolution viewed on social media) it looks like someone who did a weekend course on Live2D tried animating a still photo (but some gremlins popped by to fuck with the teeth in the animation)
@shi It looks far more realistic to me than someone who did a weekend course on Live2D.
Also, you watched it with prior knowledge that it is AI generated. My point is for those who are doomscrolling through TikTok and not applying their critical faculties.
And I know you didn't comment on possible future developments.
But I did.
And the reason I did was to point out that we should be looking at ways to deal with this kind of thing while the telltale signs are still present.