DCGAN: how do we construct the proposed CNN of the original DCGAN paper?

I am reading the original paper on the Deep Convolutional GAN (link: DCGAN paper) and on the fourth page the authors make a proposition on how to model the generator and discriminator as CNNs. However I have failed to find source code to their implementations of this construction. Their description of the layers of the CNNs are a little vague to me and not a definite recipe that I can follow.

In my research I am interested in finding an optimal construction for the DCGAN when we have images as data (maybe this goes without saying). Perhaps there are different approaches for different data sets? Where can I find a decent, or best, DCGAN available to us at this point that is already implemented (I am specifically looking for some Github repository)?

As reference, what I am looking at right now is this code: DCGAN-code.

EDIT

I found the code short while after I posted the question (typically). Here it is for anyone interested: DCGAN-sourcecode.