avatarify-python

> Correctly faking the effects of monitor illumination is a non trivial challenge for well-chosen illumination patterns. Surely not. If you have a face in a specific orientation and lighting the whole point of deep fakes is that it generates “style transfer” of the incoming image, including lighting. Doing it in real time is a computational restriction that makes it expensive, not difficult. Are you really suggesting that the deepfake models can’t accurately reproduce structured light and shadow patterns? I find that very surprising, and not what the paper says (“ We verified that the deep fakes created by Avatar- ify (github.com/alievk/avatarify-python) do not incorporate the environmental lighting and are therefore easily identifiable because in the presence of our active illu- mination, their temporal facial hue is flatlined with a nea

> I think there should be a software solution that blurs faces on the webcam Avatarify looks pretty neat too: https://www.youtube.com/watch?v=Q7LFDT-FRzs Github: https://github.com/alievk/avatarify-python

Very cool, reminds me of Avatarify, which is also based upon the First Order Model work: https://github.com/alievk/avatarify

I ran this on ubuntu 18.04. It took a little work to get around a small bug that will be squashed when the v412loopback library gets officially rebuilt but here is a solution https://github.com/alievk/avatarify/issues/37#issuecomment-6... anyway, on a 4930k at 4.5 ghz i am seeing reasonable performance(~30fps) but minimal cuda utilization(titan pascal). The biggest issue is that you need a well lit, stationary face for it to properly map features of the jpg you are using to substitute for your face. Also the jpg needs clearly identifiable features. Even then, the amount of facial expression is subdued (for example closing your eyes is not properly processed). I seem to recall software about 10 years ago where you drew line segments of corresponding features on 2 images and the jpg was then mapped onto the video. It was more accurate and

Actual Github repo for this called Avatarify: https://github.com/alievk/avatarify Original repo code that Avatarify is based on called First Order Model: https://github.com/AliaksandrSiarohin/first-order-model Short video demonstrating First Order Model: https://www.youtube.com/watch?v=mUfJOQKdtAk

Found something similar yesterday, which is basically deepfake for avatars - https://github.com/alievk/avatarify

We adapted First Order Motion Model for generating real-time avatars in Zoom/Skype. Only one image is required to build an avatar of the person. Avatarify streams generated avatars to the virtual camera. Since Zoom/Skype doesn't distinguish virtual and ordinary web cameras, you can use Avatarify camera for conferencing. At the moment, Avatarify runs on GPU (33 FPS on 1080 Ti). Supported systems are Linux and Mac OSX. CPU is supported but not yet optimized (~1 FPS on MacBook Pro 2018). The goal of this project is to make neural avatars widely accessible, so any contribution to the code is highly appreciated! We'll be glad to answer any questions in the comments. Code: https://github.com/alievk/avatarify Demo video: http://youtu.be/Q7LFDT-FRzs Fun video with Elon Musk: http://youtu.be/lONuXGNqLO0