Wow thanks for the explanation, I would they would just use what you wrote for the abstract.

At first, I thought the machine learning algorithm was able to just bridge the gap in images from nothing...

CSI time: Enhance ... enhance... enhance!