About the motion extraction model

YuDeng / Portrait-4D

Portrait4D: Learning One-Shot 4D Head Avatar Synthesis using Synthetic Data (CVPR 24); Portrait4D-v2: Pseudo Multi-View Data Creates Better 4D Head Synthesizer (ECCV 2024)

MIT License

166 stars 6 forks source link

About the motion extraction model #5

Closed FacePoluke closed 1 month ago

FacePoluke commented 2 months ago

Thank your excellent work! The driving effect of your portrait4dv2 feels better than that of portrait4d. Does portrait4dv2 also use the same motion extraction network, which is the same as PDFGC?

YuDeng commented 2 months ago

Hi, we use the same PDFGC for v1 and v2. The main differences between them are the training strategy and data. V2 uses real video data for training so its motion control is more natural.