Download network weights and place file in the avatarify directory (don't unpack it).
Clone avatarify and install its dependencies (sudo privelege is required): bashgit clone avatarifybash scripts/install.sh3. Download Miniconda Python 3.7 and install using command: bashbash Miniconda3-latest-Linux-x86_64.sh2. Linux uses v4l2loopback to create virtual camera.ġ. Of course, you also need a webcam! Install Download network weightsĭownload model's weights from here or here or here Linux There are no special PC requirements for this mode, only a stable internet connection. You can also run Avatarify remotely on Google Colab (easy) or on a dedicated server with a GPU (harder). GeForce GTX 1080 Ti: 33 frames per second.These are performance metrics for some hardware: Otherwise it will fallback to the central processor and run very slowly. To run Avatarify locally you need a CUDA-enabled (NVIDIA) video card. You can run Avatarify in two modes: locally and remotely. Every time you push the button – new avatar is sampled. Just press Q and now you drive a person that never existed. Added remote GPU support for all platforms (based on mynameisfiber's solution).
Now you can run Avatarify on any computer without GPU!
Thanks to mikaelhg and mintmaker for contribution! Now you can run Avatarify from Docker on Linux. Open Avatarify is not affiliated with Avatarify Inc.
Photorealistic avatars for video-conferencing apps. :arrow_forward: AI-generated Elon Musk Open Avatarify Avatars for Zoom, Skype and other video-conferencing apps.