Face generator powerd by AI
Each toggle below represents a point in the latent space as an input for the generator.
The latent space is a 100-dimensional input drawn from the gaussian distriubtion with a mean of 0 and standard deviation of 1. During training the generator learns to map each point to a feature of the generated object. This latent space is interpreted by the generator to create a new image.
Each point can be adjusted below to slightly alter the image.
Click on the image to randomly select a new latent space and image.
This model is a recreation of DCGAN that is described by Radford et. al. in the paper Unsupervised Representation Learning With Deep Convolutional Generative Adversarial Networks .