71–80 of 129 posts
I'm curious what would happen if somebody tried to impersonate the US president using this. He would say it's fake, but who would believe him? How exactly can computer scientists explain deepfakes to laymen?
I'm probably being captain obvious here, but if this is what's being released for free, I wonder how much better a polished commercial version does, and when we reach the point where we can't trust anything we see anymore. It doesn't even have to be super perfect, even reaching the point where it takes experts about two weeks to determine if something's real or not might already be long enough to do great damage. Fro…
Earlier quoted context omitted.
DRM will be pushed hard, starting from video/audio acquisition, perhaps assisted by blockchain to keep footage verified at all processing steps.
Don't you think that a blockchain that works for anything other than a rather useless currency should be created before suggesting one for such a use? I see comments all the time about how we should use blockchain for this and that, and yet far simpler uses for blockchain haven't yet worked out.
I'm probably being captain obvious here, but if this is what's being released for free, I wonder how much better a polished commercial version does, and when we reach the point where we can't trust anything we see anymore. It doesn't even have to be super perfect, even reaching the point where it takes experts about two weeks to determine if something's real or not might already be long enough to do great damage. Fro…
Isn’t this something a lookalike and some good make up could already do for ages?
I'm probably being captain obvious here, but if this is what's being released for free, I wonder how much better a polished commercial version does, and when we reach the point where we can't trust anything we see anymore. It doesn't even have to be super perfect, even reaching the point where it takes experts about two weeks to determine if something's real or not might already be long enough to do great damage. Fro…
The type of neural network they use (GAN) works by having two networks battling each other, one tries to generate fake, and another (discriminator) tries to identify fake, it's a constant arms race. As the generator gets better, the discriminator also gets better. Which means, if the fake video is this good, there must be a discriminator network that identifies fakes just as good.
We did a similar project using GAN, generating images from a text description. You can see the progression of generator and discriminator battling each other, and both get better with time.
They could even generate a new "cast" for each market, after only shooting the show once.
From the paper: "we have to use all the GPUs in DGX1 (8 V100 GPUs, each with 16GB memory) for training. We distribute the generator computation task to 4 GPUs and the discriminator computation task to the other 4 GPUs. Training takes ∼10 days for 2K resolution." As I don't have a DGX1 here, training the 2K resolution net for 10 days on a p3.16xlarge instance (also has 8 V100 GPUs) would cost USD 5875 on AWS. (USD24.4…
Am i reading that right? Its making the videos that look real, from the simplistic input? If so, that is amazing. And if so, how do I turn a video I have into a simple/line version, to be able to then put a different 'skin' on it?
The level of realism can be gauged from the examples they provide right there on the page. Of course your results may vary basing on the initial bulk of data of realistic source images you use. You have the code right there on Github, just install it on some PC with powerful GPUs (or rent one), tune some parameters, train the network and you can do the same things.
From the paper: "we have to use all the GPUs in DGX1 (8 V100 GPUs, each with 16GB memory) for training. We distribute the generator computation task to 4 GPUs and the discriminator computation task to the other 4 GPUs. Training takes ∼10 days for 2K resolution." As I don't have a DGX1 here, training the 2K resolution net for 10 days on a p3.16xlarge instance (also has 8 V100 GPUs) would cost USD 5875 on AWS. (USD24.4…
From the paper: "we have to use all the GPUs in DGX1 (8 V100 GPUs, each with 16GB memory) for training. We distribute the generator computation task to 4 GPUs and the discriminator computation task to the other 4 GPUs. Training takes ∼10 days for 2K resolution." As I don't have a DGX1 here, training the 2K resolution net for 10 days on a p3.16xlarge instance (also has 8 V100 GPUs) would cost USD 5875 on AWS. (USD24.4…
The DXG-1 costs $129,000 so AWS is cheaper unless you need to do it 22 times. And you can have multiple instances running at once and get all of your results in ten days, instead of waiting ten days again for each run.