generation target domain image for the given selfie image

[wzy0903]

Nice job, thanks for your open source code.
I have a question about that this code generate the target domain image from noise? or from the source data?
Is it can generate target style image from the given image? Like generation animation characters for myself, like to image-to-image translation task.