Face morphing using 3d-aware appearance optimization

Figure 1: Our system can generate fully automatically high quality face morphing animation between faces of different pose and expression. Top: morphing between images of the same subject. Bottom: morphing between different subjects. The input images are the first and last column (highlighted in red). Traditional automatic face morphing techniques tend to generate blurry intermediate frames when the two input faces differ signif-icantly. We propose a new face morphing approach that deals ex-plicitly with large pose and expression variations. We recover the 3D face geometry of the input images using a projection on a pre-learned 3D face subspace. The geometry is interpolated by factor-ing the expression and pose and varying them smoothly across the sequence. Finally we pose the morphing problem as an iterative op-timization with an objective that combines similarity of each frame to the geometry-induced warped sources, with a similarity between neighboring frames for temporal coherence. Experimental results show that our method can generate higher quality face morphing results for more extreme pose, expression and appearance changes than previous methods.