[D] What are the lessons learned in the preparations of the dataset you will use to train a GANs?

Hello friends, what are the key points we should pay attention to in the datasets you will prepare for GANs, do you have any suggestions?

For example the distribution of the dataset should be like this, the images should be the same size, it is important to reduce all the images to this size, many things that I have not thought of at the moment? What are your recommendations?

dataset gans lessons learned machinelearning