Artist Agent: A Reinforcement Learning Approach to Automatic Stroke Generation in Oriental Ink Painting

Abstract

SUMMARYOriental ink painting, called Sumi-e, is one of the most distinctive painting styles and has attracted artists around the world. Major challenges in Sumi-e simulation are to abstract complex scene information and reproduce smooth and natural brush strokes. To automatically generate such strokes, we propose to model the brush as a reinforcement learning agent, and let the agent learn the desired brush-trajectories by maximizing the sum of rewards in the policy search framework. To achieve better performance, we provide elaborate design of actions, states, and rewards specifically tailored for a Sumi-e agent. The effectiveness of our proposed approach is demonstrated through experiments on Sumi-e simulation.

Stream Flow Painting from Real Images Using Anisotropic Band-Pass Filter

“…Moreover, by defining weighting function over sectors of disc and over sectors of ellipse, anisotropic Kuwahara filter was proposed for image and video abstraction [21]. Structure adaptive filter for image abstraction is a different method which filters image in a fast separated way over an elliptic region along gradient direction and then along edge tangent direction [25].…”

“…Zhao proposed an interactive control of perceptual entropy for abstract painting from real images [23]. Painterly rendering for Sumi-e is also reported [24,25] which simulates artist's actions by using reinforcement learning to generate strokes.…”

“…As analyzed in [25], the eigenvalues provide very important information about image structures. If pixels are located in regions with low contrast, the eigenvalues are characterized by 0 …”

RealBrush

“…Our focus is a more general digital painting system that can support arbitrary natural media as supplied by the artist while maintaining high visual fidelity. Our work belongs to the class of data-driven approaches, which includes such work as modeling virtual brushes Xu et al 2004], generating brush stroke paths [Xu and Dong 2006;Xie et al 2012], and simulating pigment effects [Xu et al 2007]. Perhaps most straightforward is using scanned marks as basic drawing units (footprints), interpolated along a path to produce novel strokes [Xie et al 2011].…”

Generation of Oil-Film-Like Images by Bilateral Infra-Envelope Filter

“…Applications of nonphotorealistic rendering include technical illustrations in manufacturing and industry; water-color-like, oil-paintinglike, and India-ink-like rendering in the field of fine arts [2]- [6]; and extended computer-graphics expressions or threedimensional (3D) animation-cell-like conversions in the visual entertainment domain. Non-photorealistic rendering can be classified among methods that exploit 3D geometry data, methods that exploit two-dimensional (2D) images such as photographs, and methods that exploit pointing devices such as computer mice.…”