I'm not an expert in GAN but I heard about encoder/decoder and the fact that picture can be generated by feeding an other picture to the encoder then decoding the resulting vector.
So is it possible to add a feature to generate an image based on an image provided by the user ?