ZiqiaoPeng / EmoTalk

This is the repository for EmoTalk: Speech-Driven Emotional Disentanglement for 3D Face Animation
116 stars 4 forks source link

Details about dataset construction #8

Open zhiyiZhang1999 opened 1 year ago

zhiyiZhang1999 commented 1 year ago

Hello, thank you for the great work! It is mentioned in the paper that the construction of the blendshape datasets is achieved by a sophisticated blendshape capture method. Could you tell me the details of the blendshape capture method, like which algorithm or deep learning network is used?