Oil Style Image Generation via Fluid Simulation

Unlike photorealistic rendering striven by traditional 3D computer graphics, NPR emphasizes artistic expression, subjective mood imbuing, and downplaying unimportant information. The paper presents a new automatic method for generating oil paintings via fluid simulation technology. A paintly rendering is built up in a series of layers, drawn with large, long and curved brush strokes, and aligned to normal of fluid reference image gradients. In this method, an improved reflectance model with multiple light sources is proposed, which can represent the variability of lighting conditions on artistic painting. Additionally, a technique for simulating the hierarchical physical appearance of oil paintings under lighting is designed. The painting is rendered by bump-mapping the painting’s colors with the image’s intensity map. Finally, color transfer technique is adopted to make the color characteristic of the generated image consistent with that of some specified artistic source images. Experimental results show that the proposed method can be used efficiently to create oil paintings similar to some Van Gogh’s style from a photograph.