Have you trained the part of eyes? There is no blink in our results. How can you achieve the performance in your demos.

Hangz-nju-cuhk / Talking-Face-Generation-DAVS

Code for Talking Face Generation by Adversarially Disentangled Audio-Visual Representation (AAAI 2019)

MIT License

818 stars 173 forks source link

Have you trained the part of eyes? There is no blink in our results. How can you achieve the performance in your demos. #31

Open Claaudiaaa opened 5 years ago

Hangz-nju-cuhk commented 5 years ago

@Claaudiaaa Hi, we did not model the eyes. We actually do not know what triggers the blinking in our results. However, they do exist with certain video-driven results.