Hello everyone. If anyone has tried to finetune infinitetalk, I have finetuned it, the action can be copyed, but the charactrer consistent is not good. Things like audio layers is ok, but vision layers of wan base model is not learned. If anyone has experience, welcome to discuss. Thanks.
Hello everyone. If anyone has tried to finetune infinitetalk, I have finetuned it, the action can be copyed, but the charactrer consistent is not good. Things like audio layers is ok, but vision layers of wan base model is not learned. If anyone has experience, welcome to discuss. Thanks.