Effective Wet-in-Wet Flow Synthesis for Ink Diffusion

Wet-in-wet flow effect is a famous phenomenon in Chinese ink paintings. In this paper, we propose a new two-stage algorithm to synthesize this renowned effect. Given a reference image, in the first stage, we render the reference image using a new color ink diffusion synthesis algorithm. This physically-based algorithm explores a new and more sensitive Kubelka-Munk (SK-M) equation. As a result, this new algorithm produces an ink diffused image, which has the similar tone to the reference image and offers better visual plausibility than our previous work [1]. In the second stage, we present a controllable flow effect technique in order to synthesize the wet-in-wet flow effect. In particular, the adaptive length line integral convolution is adopted to represent the global flow of the reference image. Given this global flow and luminance of the reference image, a controllable flow map is generated using the desired weight coefficients controlled by the user. Finally, we blend this controllable flow map with the ink diffused image rendered in the first stage. The blending takes advantage of the new SK-M equation again, synthesizing a Chinese ink diffused image with a notable wet-in-wet effect. Our algorithm has four advantages: visually pleasing, controllable, independent, and simple.