Another questions about the loss function

[qinhaifangpku]

Hi~ I have noticed that you use the sigmoid cross entropy loss function to learning the autoencoder. But if my data is a RGB image, It is impossible to use this loss. I wonder if my opinion is true?