Neat. Although in most of the examples I looked at, the NN seems to get the shape of the eyes uncannily wrong. I would have thought that would be the easy part.

Neat. Although in most of the examples I looked at, the NN seems to get the shape of the eyes uncannily wrong. I would have thought that would be the easy part. I haven't looked at the paper. I thought that this was going to be about 'Gaze Manipulation' in the sense of manipulating which part of an image draws your attention, e.g. [0] [0] https://sites.google.com/a/g.rit.edu/gaze-manipulation/ https://sites.google.com/a/g.rit.edu/gaze-manipulation/