UniFaceGAN: A Unified Framework for Temporally Consistent Facial Video Editing

Recent research has witnessed advances in facial image editing tasks including face swapping and reenactment. However, these methods are confined to dealing with one specific task at a time. In addition, for video editing, previous either simply apply transformations frame by or utilize multiple frames concatenated iterative fashion, which leads noticeable visual flickers. this paper, we propose unified temporally consistent framework termed UniFaceGAN. Based on 3D reconstruction model simple yet efficient dynamic training sample selection mechanism, our is designed handle reenactment simultaneously. To enforce the temporal consistency, novel loss constraint introduced based barycentric coordinate interpolation. Besides, region-aware conditional normalization layer replace traditional AdaIN SPADE synthesize more context-harmonious results. Compared state-of-the-art methods, generates portraits that photo-realistic smooth.