Putting your skills into practice

Now, you can apply the skills you have learned to implement your own image generation projects. Before you start, there are some pitfalls you should look out for and also some practical advice that you can follow.

Don't trust everything you read

A new academic paper is published and shows astonishing images generated by their model! Take it with a pinch of salt. Usually, these papers handpick the best result to showcase and hide the failed examples. Furthermore, the images are shrunk down to fit onto the paper, thus the image artifacts may not be visible from the paper. Before investing your time in using or re-implementing the information in the paper, try to find other resources of the claimed results. This can be the author's website or GitHub repository, which may contain the raw, high-definition images and videos.

How big is your GPU?

Deep learning models, especially GANs, are computationally expensive. Many of the state...