Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing

nvidia-research-mingyuliu.com

Post reply on HN
Print view View original

31–40 of 52 posts

Re: One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing

#31

Post by dillondoyle »

koilke wrote: post #22

Inviting terms as usual: When you upload, submit, store, send or receive User Content to or through the NVIDIA Research AI Playground, you give NVIDIA (and parties NVIDIA works with, including its affiliates, suppliers and customers) a worldwide license to use (including without limitation for neural network training), host, store, reproduce, modify, create derivative works (such as those resulting from translations,…

It doesn't sound like they are trying to sell or somehow leverage to make profit? I guess 'develop new NVIDIA offerings' is kind of vaguely suspicious but the rest doesn't seem onerous to me unless I'm missing something?

Re: One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing

#32

Post by kadoban »

dillondoyle wrote: post #31
koilke wrote: post #22

Inviting terms as usual: When you upload, submit, store, send or receive User Content to or through the NVIDIA Research AI Playground, you give NVIDIA (and parties NVIDIA works with, including its affiliates, suppliers and customers) a worldwide license to use (including without limitation for neural network training), host, store, reproduce, modify, create derivative works (such as those resulting from translations,…

It doesn't sound like they are trying to sell or somehow leverage to make profit? I guess 'develop new NVIDIA offerings' is kind of vaguely suspicious but the rest doesn't seem onerous to me unless I'm missing something?

Those license terms seem to me like they'd allow using someone's picture in unlimited advertising, for free. They also sound like they'll definitely be using them to train models. I wouldn't be surprised if some of that means they can sell them to advertisers for whatever nonsense.

Re: One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing

#33

Post by batch12 »

dheera wrote: post #28
batch12 wrote: post #25

Earlier quoted context omitted.

Would that then just be a proxy for not being a white male and present the same problems? I am not sure what this solves to prevent discrimination.

It doesn't prevent discrimination, it just allows me to do what I want to do in the short term, e.g. raise funding for startups or get dream jobs or whatever. I mean if a VC hands me a term sheet or a hiring manager hands me a job and the ONLY thing I misrepresented is my face, I don't think they have any ethical grounds to retract their offer. Solving the discrimination problem is another matter, and will take years…

Good luck to you. I am not convinced people would be more likely to give money or a job to someone when they have only seen an avatar, but what do I know.

Re: One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing

#34

Post by dheera »

batch12 wrote: post #33
dheera wrote: post #28

Earlier quoted context omitted.

It doesn't prevent discrimination, it just allows me to do what I want to do in the short term, e.g. raise funding for startups or get dream jobs or whatever. I mean if a VC hands me a term sheet or a hiring manager hands me a job and the ONLY thing I misrepresented is my face, I don't think they have any ethical grounds to retract their offer. Solving the discrimination problem is another matter, and will take years…

Good luck to you. I am not convinced people would be more likely to give money or a job to someone when they have only seen an avatar, but what do I know.

They don't have to know it's an avatar. For most jobs if they make any decision based on how the candidate looks they are in the wrong.

It's not any different from make-up, really, which is also hiding your real face, in a different way.

Re: One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing

#35

Post by batch12 »

dheera wrote: post #34
batch12 wrote: post #33

Earlier quoted context omitted.

Good luck to you. I am not convinced people would be more likely to give money or a job to someone when they have only seen an avatar, but what do I know.

They don't have to know it's an avatar. For most jobs if they make any decision based on how the candidate looks they are in the wrong. It's not any different from make-up, really, which is also hiding your real face, in a different way.

If you deceive someone by wearing a different face during an interview, I expect that you would not be hired. How do you prove you were the person interviewed?

Re: One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing

#36

Post by dheera »

batch12 wrote: post #35
dheera wrote: post #34

Earlier quoted context omitted.

They don't have to know it's an avatar. For most jobs if they make any decision based on how the candidate looks they are in the wrong. It's not any different from make-up, really, which is also hiding your real face, in a different way.

If you deceive someone by wearing a different face during an interview, I expect that you would not be hired. How do you prove you were the person interviewed?

There are plenty of ways to prove that, including stating a public key during the interview, interviewer signs it with their key, vice versa is done with the interviewer's key, and then the same is demonstrated in person with the same keys.

Hell, how to do that can even be an interview question.

Re: One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing

#37

Post by villgax »

thebruce87m wrote: post #9
kevingadd wrote: post #2

Surprisingly flexible algorithm. I threw a headshot illustration of a comic villain into it and it was able to successfully do all the rotations and even manipulate the eyes despite it not being photorealistic.

There was an app going around a few months ago called “wombo.ai” that will make a headshot sing. It had the same outcome when given non human pictures as a source. Easily killed an afternoon trying out different things.

That was using something similar to First Order Motion Model & not the same model as what Nvidia has been using for Maxine or subsequent improvements to it as in the linked demo

Re: One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing

#38

Post by forgingahead »

dheera wrote: post #28
batch12 wrote: post #25

Earlier quoted context omitted.

Would that then just be a proxy for not being a white male and present the same problems? I am not sure what this solves to prevent discrimination.

It doesn't prevent discrimination, it just allows me to do what I want to do in the short term, e.g. raise funding for startups or get dream jobs or whatever. I mean if a VC hands me a term sheet or a hiring manager hands me a job and the ONLY thing I misrepresented is my face, I don't think they have any ethical grounds to retract their offer. Solving the discrimination problem is another matter, and will take years…

Sincerely, I'm sorry you feel that you haven't achieved certain things because of skin colour/accent/etc. However let me challenge that belief, only because if your assumption isn't true, that your lack of desired success is because of things outside of your control, then perhaps you won't make the adjustments potentially needed to actually achieve them. If your dreams truly cannot wait (as you said, and they shouldn't!), then imaginary AI solutions to not-catastrophic-not-life-ending barriers shouldn't be seen as anything relevant to your eventual success.

None of us can control skin colour, accent, connected families, or other benefits that others may have. But we can outwork the privileged, and today's world, with just a computer and internet connectivity, gives us all much more options than previously imaginable.

I do sincerely wish you great luck and success, but I really think you shouldn't get stuck on things you can't control, and devote all your energy to things you can.

Re: One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing

#39

Post by sb057 »

koilke wrote: post #22

Inviting terms as usual: When you upload, submit, store, send or receive User Content to or through the NVIDIA Research AI Playground, you give NVIDIA (and parties NVIDIA works with, including its affiliates, suppliers and customers) a worldwide license to use (including without limitation for neural network training), host, store, reproduce, modify, create derivative works (such as those resulting from translations,…

Sounds like a general CYA license so that developers can try it out before an actual product is rolled out with a proper commercial license.

Re: One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing

#40

Post by fulafel »

knicholes wrote: post #6

I'm getting connection refused errors on the request to http://54.186.34.220:443/face_vid2vid_rotate port 443 on http? Hm.

Interesting that they're running plaintext HTTP on the HTTPS port. Are some networks filtering 80 on egress now and this is how people get around it? Or are the devs just using some cloud setup that tries to force serving HTTPS but the builders of the service don't want to?
Post reply on HN