New developments in the field of AI offer impressive capabilities to generate lifelike images. But how do these work, what are the limitations, and in what products we might see them in the future?