Some time ago, a new neural algorithm came out (detailed in this paper) which allows people to transform a picture/painting/image using the style of another image (I used images of paintings, mostly).
The code is available there:
The link to the gallery.
Here are a few images I have made so far:
Basically, the algorithm decomposes each of the two images (the style and the content) into layers of information (shapes, colors, lines, etc.) and then tries to apply each layer of the style to the corresponding layer of the content. It is a very intensive computation : about 1000s for a 512×512 on the CPU or 10s on the GPU, however unless you have a graphic card with lots of memory you will not be able to create very high resolution images (mine crashed over 600×600 for lack of RAM).