Figure 1. Given an input image (a) and a given reference face image, our method can independently edit the geometry of any local region, such as hair, nose, mouth, eyebrows, etc. We show these local and global face editing tasks achieved with our method: (b) editing the geometry of the hair by modifying the semantic mask while retaining the geometry of other regions and 3D consistency; (c) editing the geometry of more local regions, such as nose, eyebrows and mouth; (d) editing the local texture of the hair while retaining the geometry and 3D consistency; (e) editing the global texture while retaining the geometry and 3D consistency.