| The stylized painting of images or models has become a hot topic in the realm of Non-Photorealistic Rendering, for it can simulate the processes and effects of different kind art styles based on the image identifying and analysis with the help of artistic ideas and artistic standards of artistic creation. In this paper, we regard images and models as the entry points of our research. We use the method of style transfer based on brushstroke synthesis, and perform on systematic research on two key problems: style transfer of 2D images and style transfer of 3D models. Then we develop the prototype system named Image Palette base on the previous researches. Generally, the main contributions of this paper are:1. Style Transfer of 2D Images Based on Brushstroke Synthesis. For sample im-ages with certain styles and a target image, in our approach, users can take any input sample images with a certain style as a palette according to their preferences, and then perform personalized art style transfer on the target images via the simple se-lection of arbitrary contour or texture as each stroke input mode. We then perform on style transfer synthesis along the stroke curves learned from the specified area in a target image to get the same painting style as the samples. The experiment result shows that, in our approach, we can accomplish style transfer of images efficiently, and lay a solid foundation for subsequent style transfer of models.2. Style Transfer of 3D models Based on Brushstroke Synthesis. For sample im-ages with certain styles and a target 3D model, we unwrap surface of 3D models according the user’s intent to obtain 2D unwrapped images. Then we let users take any input sample images with a certain style as a palette, and perform personalized art style transfer on 2D unwrapped images via the simple selection of arbitrary con-tour or texture as each stroke input mode. The experiment result shows that, in our approach, we can accomplish style transfer of from sample images to 3D models according to users’intent.3. Application system of Image Palette. We combine the previous researches into the application system. Our system can make the process of style transfer quick and efficient with user-friendly interactions to help users complete style transfer effectively. |
