A generative model for faces based on Conditional VAE-GAN. Demo – M. Suzuki