You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@Olivia-Yeclip_path specifies ground truth which is where face keypoints are stored. the model outputs upper body configuration, exclusive of face landmarks.
问题包括以下方面:
1、new speaker的keypoint是怎么生成的?
2、我想要生成新的人物演讲动作,我需要从该人物的以往手势视频中获取lab、wav、textgrid(通过mfa音素对齐),此外还要进行openpose获取keypoint吗?
3、请问您提供的预训练模型是通用的吗?(我的意思是指:不需要指定某个说话人,生成他的动作风格)
如果您能提供一个联系方式来讨论问题,不胜感激!我已经为您这个项目忙活了半个月了,实在是很多地方不太理解。
The text was updated successfully, but these errors were encountered: