Expression transfer

Since 3D landmark output is possible, would it be possible to transfer the expressions from one tracked face to another while keeping the original head position (like [this]( https://youtu.be/ohmajJTcpNk?t=73) but with the landmarks)?

The issue references 3D landmark output and expression transfer. Start by examining the 3D landmark extraction code in the face-alignment library, likely in the core alignment modules. Review the linked video to understand the desired expression transfer effect. Check if there are existing functions for landmark manipulation or expression modeling. Determine how to separate expression parameters from head pose using the 3D landmarks.

Written by the indexing model from the issue text.