Viewing profile — huling0

though I have to admit this looks like the best in class, there is a public real time deep fake service already available: https://www.aphrodite.ai/

http://notarealhuman.com/ has cats as well :)

good suggestion. Looking into latent vector manipulation now.

Why this negativity? It's not harming anyone. Just fun to play with the new StyleGAN network which is only out since a few weeks.

Thanks! I missed that one - seems to do the same :-)

It is based on https://arxiv.org/pdf/1812.04948.pdf Images can easily be generated on the fly. Inference time is ~100ms.

Care to share what weekly subscription newspaper you read?

I recently ran into that issue - I have a nice fine-tuned yocto image and now a customer wants to "log in via ssh and install a custom raspbian package". I was thinking of the dock…

yeah, but does not look like something many hobbyists would try for fun ;-)

Yes, but it just cuts out the face and pastes it on a different person/background. It does not do full reenactment where you keep the entire target video environment.

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…

One of the example translates a full human pose to a video of a dancer. If the network would be trained on the facial pose(?) / features only, would that recreate something like th…