A bit more info here (source)
So this academic Deep Learning System (DLS) can take an input photo, analyse a reference photo and apply the art style to produce a version of the input photo in the style of the reference photo.
So could something similar be applied to games, could an DLS make your game look more like another game, photo, painting or even your artists original themed paintings?
Or what impact could this have for game artists?
Seems a bit random though, erasing all and any need for a human’s memories and experiences going into art. The point of making art is that it’s a human expression, right?
Art style is an enjoyable bit of dev for us, we don’t want to automate it.
Yeah I’ve got to wonder what use this would be. I can see a need for something that helps to create background art in the same style as something you’ve already done, so as to reduce workload. But something that simply transforms your hard work into someone else’s style, doesn’t seem like an interesting thing to me. Style is an end in itself, it’s not just a coat of paint.
It would make brilliant artwork to signify the process of encroaching insanity, a peak experience mystic state or ingestion of mind altering substances…states that are very hard to convey virtually but would make great additions to magickal based RPG’s, portal jumpers. It would probably also be able to cobble up some wild VR scenarios and landscapes, portraitures, morphanimations.
Something like this would most likely look horrifying in motion.
Perfect for 'Apocalyptic Game Style’ 
It’s not a what if
It’s happening and it’s just one more tools
But I do expect neural network to generate movie out of thin air in like 5 years, complete with voice. In 10 years it will be fully mature.
BTW I was following @dribnet on twitter and its experiment is now a app.
The question is not what if, it will change everything. AAA are already really seriously into it, in fact on the forefront of experimentation there is Alex J Champaniard, who worked on killzone and has the website AI gamedev that have in depth talk with many industry leader of the field.
I would say indies would benefit the most from going into it early, but for some reason we are seriously lagging. Indies has become the risk adverse party now.