I find it unbelieveable how some advanced neural networks can take source face and transpose it to target face with almost 100% precision even in facial expressions - all from one picture. If you told me that would be possible 10 years ago, i would tell you you read too much scifi.

So how do they work exactly?