The architecture of a 3D-GAN

Both of the networks in a 3D-GAN are deep convolutional neural networks. The generator network is, as usual, an upsampling network. It upsamples a noise vector (a vector from probabilistic latent space) to generate a 3D image with a shape that is similar to the input image in terms of its length, breadth, height, and channels. The discriminator network is a downsampling network. Using a series of 3D convolution operations and a dense layer, it identifies whether the input data provided to it is real or fake.

In the next two sections, we will go through the architecture of the generator and the discriminator network.