How is the target frame determined

[tangyibo123]

@lelechen63 @rainingDesert Thank you very much for your project,I saw the last part of the paper，“the controllable results. The videos in upper part are manipulated with target facial expressions while keep the head motion unchanged. We show the target facial expression in the first row. The videos in lower part are manipulated with target head motion while keep the facial expression unchanged.”

But when I tested, I only input the audio and a series of pre-processing results related to the original video. How is the target frame determined?

[rainingDesert]

@lelechen63 @rainingDesert Thank you very much for your project,I saw the last part of the paper，“the controllable results. The videos in upper part are manipulated with target facial expressions while keep the head motion unchanged. We show the target facial expression in the first row. The videos in lower part are manipulated with target head motion while keep the facial expression unchanged.”

But when I tested, I only input the audio and a series of pre-processing results related to the original video. How is the target frame determined?

The head motion of target frame is determined by the head motion of original video. The facial expression is determined by landmarks synthesized according to audio, which belongs to the target and contains related information.