@DmitryUlyanov first, thanks for your great work which is really helpful to my research. I have a question after reading the paper. To apply your method to style transfer, do we need to input content image during training phase? If not, the content loss is generated for what? Based on my understanding, the content loss is used for preserving the spatial structure of the content image. But there is no content image option in train.lua.
@DmitryUlyanov first, thanks for your great work which is really helpful to my research. I have a question after reading the paper. To apply your method to style transfer, do we need to input content image during training phase? If not, the content loss is generated for what? Based on my understanding, the content loss is used for preserving the spatial structure of the content image. But there is no content image option in train.lua.