beyonehan

> 我复现了一般Keras版本，有需要的可留下邮箱，谢谢～～ 请问效果怎么样啊

> 你好，感谢你的关注。 是的，我们添加了一个人像检测分支，以尽可能地避免无人像图像输出噪声。 多谢分享， 请问模型上面的改进 就是做了这个吗。请问那个商用版效果是用了多少张图做训练啊

> I run it on 3090 but the FPS can't reach 25fps Am I missing some details? Looking forward to your response Hi, could you tell me your FPS on...

> 您好， 尝试了下你的代码，生成的20张动漫图看着很不错。看代码是先random了一个输入特征。 > > ```python > latents = np.random.randn(1, Gs.input_shape[1]) > ``` > > 想问下能否根据给定的人脸来生成特定的动漫图呢～ 同问

> 您好， 尝试了下你的代码，生成的20张动漫图看着很不错。看代码是先random了一个输入特征。 > > ```python > latents = np.random.randn(1, Gs.input_shape[1]) > ``` > > 想问下能否根据给定的人脸来生成特定的动漫图呢～ Hi，请问你有什么进展吗

I download it ,but it shows fail to load checkpoint ,can you tell me the reason  

thanks for you reply,but it still has the same error. I tried many times ,but I can find it out  

> What version of Tensorflow do you have? It requires 1.15 > > Can you run this command: > `cat UGATIT.py | grep UGATIT.model-2044000` > and provide the result '1.15.0'...

thank you very much .finally ,I got it. But when I test it with other head portrait, it seems not well.Could you tell how to get the effect like baiduAI...

> > > I have trained a model based on UGATIT-pytorch framework which is more robust to various faces and more friendly to boys' faces > > > > >...