Summarizing Most Popular Text-to-Image Synthesis Methods With Python

Embedding as illustrated in the below diagram is passed through the Conditioning Augmentation block (a single linear layer) to obtain the textual part of the latent vector (uses VAE like parameterization technique) for the GAN as input. The second part of the latent vector is random Gaussian noise. The latent vector yielded is then fed to the generator part of the GAN. The embedding thus formed is finally fed to the final layer of the discriminator for conditional distribution matching.