Closed foocker closed 2 years ago
Please check issue #33.
谢谢,还没仔细看论文,这个是一个人物,一套模型? 如果不是,能讲解下数据制作?
Yes, the proposed method is person-specific. For data processing, please check the paper.
Yes, the proposed method is person-specific. For data processing, please check the paper.
谢谢,总结了下,还是有很多细节。有点好奇,看了下历史issue,既然这么多人复现了,怎么没有一个pr的?
@YuanxunLu 训练数据的目录结构方便发一下吗?h5里面的数据是?change_paras中的scale, xc, yc咋计算的?
Releasing the training data & structure is not in the scope of this repo for copyright reasons, check issue #33.
The released code, including training settings, dataset files, etc will help you to train your own model.
Releasing the training data & structure is not in the scope of this repo for copyright reasons, check issue #33.
The released code, including training settings, dataset files, etc will help you to train your own model.
ok,我重新开了一个问题,希望您可以选择性的回答一些疑问。
我训练出来了,比LiveSpeechPortrait的画面更稳定,演示效果 https://space.bilibili.com/3494353091693235
@ak01user I am facing difficulty in implementing this. can you please share some of your training code?
Please check issue #33.