Ask HN: What are some bleeding edge Generative Neural Network architectures?

I am about to start experimenting with making my own architectures for body pose estimation from color images (ala OpenPose) and if I'm not mistaken, the problem can be solved either via generative (GANs, VAEs) or discriminative networks (CNNs). So far I prefer generative networks over discriminative ones.

As the title says, I am curious what are the cutting edge generative network architectures. Are there any other architectures besides GANs and VAEs? Where can I learn more about this?