We can generate images just based on the text using an interesting type of GAN called StackGAN. The StackGAN works in two stages:
In the first stage, they generate a basic outline, primitive shapes, and create a low-resolution version of the image.
In the second stage, they enhance the picture generated in the first stage by making it more realistic and then convert them into a high-resolution image.
In the Cycle GAN, the generator map the images from the source domain to a random permutation of images in the target domain which can match the target distribution. So, to alleviate this, we use a special type of loss called cycle consistent loss.
Cycle GAN is majorly used in the premises where it is difficult to obtain the paired training samples. Some of the interesting applications of CycleGAN include photo enhancement, season transfer, converting the real pictures to the artistic pictures, and so on.
We can perceive InfoGAN as an unsupervised version of conditional GAN. In the conditional GAN, we impose a condition on the generator and discriminator to generate the image we want based on the class labels present in the dataset. When we have the unlabelled dataset then we can use InfoGAN for generating the images we want.