very very cool. One thing I noticed that with image to image translation tasks the output tends to be a bit "organic" looking, like the photo to map example. Wi

very very cool. One thing I noticed that with image to image translation tasks the output tends to be a bit "organic" looking, like the photo to map example. With photographic output it's not noticeable, but it's very jarring for graphical output. I wonder if there is a way to fix this, possibly by stacking another GAN on top?