Plea for Help about Gaussian Copulas

I am not ashamed to admit that even though I understand the concept of Gaussian copulas a lot better than I did six months ago, the information has not gelled together into a cohesive whole in my mind.

From what I've read, let's say x1 is a distribution that isn't normal and x2 is a distribution that isn't normal either. Our goal is to find the correlation between these two distributions, but this can only be done indirectly by relating each of these distributions to two standard normal distributions.

I'm getting most of this from John Hull's book, 'Risk Management and Financial Institutions.' Hull provides the most coherent explanation of everything I've read.

The correlation of the two transformed distributions is apparently the Gaussian copola. The copula is then used to measure default risk.

I don't understand how the conversions are made and what they mean.

I'm not asking about this because I want to use copulas to make money in the market. I realize that David X Li's formula for the Gaussian copula has been discredited. Correlations are not constant; they change all the time. And the copulas that Li and others were using were based on credit default swaps before the bubble burst, so the data was from a small sample with market conditions that changed very quickly.

I'm trying to understand copulas better for this weekend's Part 2 FRM exam.

Any and all insights from the WSO community will be more than welcome!