flysky126

Results 5 issues of flysky126

Hi, I train with vox-adv version and set sn=true, after train 4 epoch, the infernce result is strange, have you happen to this? ![1643339334](https://user-images.githubusercontent.com/71961564/151480663-e30138e0-d9ef-4031-a09c-9bb423f05f7a.png)

Hello @YuvalNirkin ,I train model using own dataset, but blending result is below, Is the reason that model not train well or train data not enough? ![20201021090149](https://user-images.githubusercontent.com/71961564/96660502-476b0300-137c-11eb-8978-4c1adfee780d.jpg) Thank you very...

![image](https://github.com/ZiqiaoPeng/SyncTalk/assets/71961564/799c649e-b7c5-4010-94b8-7b8ee784f372) 请问各位数据预处理中这个mask是指头部的哪个区域了? 包括头发和脖子区域吗?

训练了一个新的id,生成视频的头部运动频率跟原视频不同,有抖动,导致生成结果贴回原退去时会有断层,请问有啥方法可以减小生成视频的头部抖动频率吗?

由于输入是有一张masked图和模板原图组成,在生成修复帧的时候,模板图的动作会影响到生成的修复帧(例如模板是张开嘴部的,但是修复帧是要嘴巴闭合,这样就造成下巴会向下拉扯),请教下大家这个问题了。。。