Abstract: A traditional digital oil painting canvas consists of lines that represent region boundaries and
digits that represent specified pigments. It becomes a new popular art because amateurish users can
easily draw their own oil painting by brushing provided pigments into each region. However, it costs
professional artists much labor and time to design, and reduces the painting quality due to the lack of
brush hints, such as their orientation and width. In this paper, we propose an automatic generation
method with brush hints. It not only converts an arbitrary image into a digital oil painting canvas, but
also provides effective brush hints on the canvas. Experiments demonstrate the smoothness of boundary
lines, the legibility of color digits and the significant improvement of painting quality contributed by the
hint of brush streamlines.

Key words: digital oil painting, image segmentation, field streamline, line extraction, non-photorealistic
rendering