Abstract
The article presents an algorithm for texture transfer between images that is up to several orders of magnitude faster than current state-of-the-art techniques. I demonstrate how the technique can leverage self-similarity of complex images to increase resolution of some types of images and to create novel, artistic looking images from photographs without any prior artistic source. Compared to other alternatives, methods based on texture transfer are global in the sense that the user need not deal with details such as defining and painting individual brush strokes. Texture transfer methods are also more general since they don't need to emulate any particular artistic style (line drawing, hatching, realistic oil painting, and so on). Not surprisingly, there is a price to pay for this generality - an algorithm designed for a specific artistic style will most likely produce results superior to those presented in the paper for that particular case.

This publication has 8 references indexed in Scilit: