Choice between multi-scale discriminator and progrssive growing of GANs ?

[leo27945875]

Hi,

I saw that GauGAN uses multi-scale discriminator just like pix2pixHD in the paper.
I'm wondering that have you ever considered to use the training trick in ProGAN, which is also proposed by Nvidia ?
Because the concepts between them are similar (I think).

Thanks !