FaceController: Controllable Attribute Editing for Face in the Wild

Face attribute editing aims to generate faces with one or multiple desired face attributes manipulated while other details are preserved. Unlike prior works such as GAN inversion which has an expensive reverse mapping process, we propose a simple feed-forward network high-fidelity faces. By simply employing some existing and easy-obtainable information, our method can control, transfer, edit diverse of in the wild. The proposed consequently be applied various applications swapping, relighting, makeup transfer. In method, decouple identity, expression, pose, illumination by using 3D priors; separate texture colors region-wise style codes. All information is embedded into adversarial learning identity-style normalization module. Disentanglement losses enhance generator extract independently from each attribute. Comprehensive quantitative qualitative evaluations have been conducted. single framework, achieves best competitive scores on variety applications.