Nvidia unveils new tech. to turn quick doodles into photoreal images

This is not clickbait, Nvidia actually created a neural network where you can do some crappy sketch and an algorithm turns it into something like a photo.

A couple of limitations though, it looks like the algorithm gets an assist by requiring the ‘artist’ to use certain colors for certain elements, and it also seems to be very focused on landscapes. In addition, this is another one of those things that you can’t download and use at the moment.

Still though, there is a good use for this as something to turn an idea sketch into a reference for something like a game environment or CG art. In my opinion, this is honestly more impressive than previous neural stuff since it explicitly shows user art as the input and how it transforms it in real-time.