Workshop on Computational Photography and Aesthetics

This talk will discuss several challenges in non-photorealistic image rendering.

Emulating the art of the classic engravers, like Durer, requires the development of algorithms to generate curve networks whose local density and flow is intelligently controlled by the image and its gradient, and this can be done by level-set based shape from shading techniques. Picasso, or Calder-like renderings require procedures to extract essential contours, or edges from grey-level images and connect them in interesting, naturally flowing ways, and this leads to some new and interesting mathematical challenges.

The presentation will be based on several papers deaing with non-photorealistic, engraving-like renderings, and with image representations by one-liners, describing our (hopeless!) attempts to emulate grand masters like Durer, Picasso and Calder.

J. Rigau, M. Feixas, and M. Sbert, Image information in digital photography

P.-Y. Lo, S.-W. Shih, J.-C. Liu, and J.-S. Hong, Automatically detecting protruding objects when shooting environmental portraits

T. Sachs, R. Kakarala, S. Castleman, and D. Rajan, A data-driven approach to understanding skill in photographic composition

M. Constable, Artist-led suggestions towards an approach in content aware 3D NPR rendering