An illustration of a magnifying glass. An illustration of a magnifying glass.

Abstract: Encoder-decoder based architecture has been widely used in the generator of generative adversarial networks for facial manipulation. However, we observe that the current architecture fails ...