Paintbrush rendering of lines using HMMs
Author(s): Yuta Okabe, Suguru Saito, Masayuki Nakajima.
Proceedings: 3rd International Conference on Computer Graphics and Interactive Techniques in Australasia and South East Asia (GRAPHITE'05), pp. 91--98, Dunedin, New Zealand,
2005.
[BibTeX] [DOI] 
Abstract:
We propose a new method that emulates the expressive appearance of paintbrush strokes on paper. In our system, brush footprint characteristics are modeled using Hidden Markov Models (HMMs). Once the model is trained by a data set of given strokes, it estimates the most likely sequence of footprints for a given input line. The main difference between our method and other ones that use direct texture mapping lies in the estimation of the dynamic behavior of the footprints with respect to input lines. Since our method models simultaneously the physical characteristics of the brush and the drawing styles, it transforms input lines to aesthetic strokes which are really close to true ones, made with a real brush on a paper.