Disentangled and Controllable Face Image Generation via 3D Imitative-Contrastive Learning (CVPR 2020 Oral) - View it on GitHub