Real-Time Face Capture And Reenactment Of RGB Videos

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
Remember the real-time facial reenactment video we showed you awhile back? Well, those guys are at it again and they've cranked up the realism. If you haven't seen this is action yet, you really need to watch this video.
 
I don't know about you, but that's just damned creepy. It won't take long before professionals take over for the politicians, and we won't know any better.
 
I am not sure to think that is cool or really scary. Now you can easily make a fake video of somebody saying thing bad. Yes you would still need to get a good audio spoof of them saying something, but with the video to match it up to that wouldn't be the hardest part.
 
What a better time to be dead, and live again! Kids can have bedroom stories read to them by their great great grandparents. Alec Guinness can be in another star wars episode.

Or people will just use this tech to leak fake libelous videos of our neighbors and enemies. Soon we can't even rely on video as evidence since it's so easy to taint.
 
I don't know about you, but that's just damned creepy. It won't take long before professionals take over for the politicians, and we won't know any better.

Politicians already don't speak for themselves....they have a team. This just takes it the next logical step further.
 
want to have some fun try msqrd on Iphone or Android msqrd.me
 
Politicians already don't speak for themselves....they have a team. This just takes it the next logical step further.
They'll just datamine how people actually speak sentences and move their faces, improve voice synthesis, and then replace the speech writers with computers that datamine what people want to hear from your political candidates. Eventually we can vote for the democratic national AI or the republican national AI. Humans need not apply. We'll have a Hatsune Miku like president that accurately reflects the political policies and will of congress as a whole. Every four years, they'll find the correct facial and voice model to pitch as the face of the president.
 
I don't see this as anything good. If you want to parody someone it should be transparent.
This will be used to spoof such things as news reports . YouTube will be flooded with take-down notices, but in most cases the damage will be done the second it goes viral.
 
I don't see this as anything good. If you want to parody someone it should be transparent.
This will be used to spoof such things as news reports . YouTube will be flooded with take-down notices, but in most cases the damage will be done the second it goes viral.

It could lower the bar for making video games where the facial animations are more realistic. Indy film makers would probably love this for postproduction since it doesn't require expensive hardware and is real time.
 
You know, ALL I WANT is to have ~cheap access to this software to make game animations. But NO. NOBODY will allow this.
 
This is pretty scary, this will be on every cell phone in 10 years I'm sure.
 
I don't know about you, but that's just damned creepy. It won't take long before professionals take over for the politicians, and we won't know any better.
Sadly, most politicians are just talking head puppets anyway. As soon as one is never seen in public, and only makes teleconference style appearances, we'll know what's really going on.
 
"Pay no attention to the man behind the curtain.", comes to mind.

What the heck is this supposed to be good for anyway?
 
I am not sure to think that is cool or really scary. Now you can easily make a fake video of somebody saying thing bad. Yes you would still need to get a good audio spoof of them saying something, but with the video to match it up to that wouldn't be the hardest part.

Actually, with enough audio samples, the technology already exists to build a voice synthesizer/emulator (forget what the actual term is, but it's an acoustic model of the person's speech, and it works pretty well with arbitrary input).
 
Do they realize how disturbing it is to watch words put in the mouths of major world leaders in a news broadcast format in real time? Gee, what could possibly go wrong?
 
Actually, with enough audio samples, the technology already exists to build a voice synthesizer/emulator (forget what the actual term is, but it's an acoustic model of the person's speech, and it works pretty well with arbitrary input).

Yes. I am sorry I was trying to say in my post that between making a fake video and making fake audio. The audio part wouldn't be that hard and now they make the fake video part easy. Meaning fake videos are not going to be that hard at all now.
 
Actually, with enough audio samples, the technology already exists to build a voice synthesizer/emulator (forget what the actual term is, but it's an acoustic model of the person's speech, and it works pretty well with arbitrary input).
Is it that good already? I didn't think that the psychoacoustic models were that good, since it's more difficult to train a computer to match contextual changes to intonation which you get with arbitrary input. Most voice modification or synthesis I've heard still sounds unnatural, sounding flat at times where there should be contextual anger, excitement, boredom, etc. (I'd love to know what the cutting edge is right now).
 
Wow, this is cool yet kinda scary. If it gets to the point where it is perfect/near perfect, imagine a political candidate on a live interview being asked softball questions, while the interviewer and in the interviewee are replaced by off screen actors asking different questions and giving different answers.
 
Too bad everyone immediately thinks of political usage. I want games to be animated like this. Also you could choose how you want your news anchor to look on the fly, that would be cool.
 
This is scary. For so many people; reality=what is on TV.
We are fast approaching a time where nothing we are shown as "the truth" is true at all.
 
Back
Top