GHARP emphasizes that only a few input images are needed, which can lower the data collection barrier, and users may only need a few multi-angle selfies to obtain a 3D head that can be driven in real time.