speech driven avatar

1. Control by FLAME expression parameter

   Given New person face (video or image)

   • FLAME Shape paramters
     • MICA : Texture 코드 추가
     • [ ] DECA

   Control By FLAME expression parameters

   • Emotalk benshapes → blenshape2FLAME → FLame expression parameters
   • [ ] blendshape2FLAME (by scan FLAME fitting)

2. Control by blendshapes

   Input (Neutral) Image → MICA (Mesh, Texutres, Landmarks) → LDT (Transfer to human specific blendshapes) → EmoTalk Direct Control

   • Todo
     1. MICA : Texture 코드 추가, Face Landmark → Vertex Landmark (오늘)
     2. LDT : GUI 베이스 코드인데 → Pipeline에 연결하기 (내일)

3. Unity 통신