Image palette: painting style transfer via brushstroke control synthesis

摘要

As one kind of technology of style transfer, painting style transfer can be used to render the sample images with a specific art style. With this technology, we can render the target images in the same style as the samples after some computation. We present a new approach of painting style transfer in which such a style transfer artwork is done by simulating the process of creation. We take the sample as a palette where users can select arbitrary contours or textures as the input brush strokes. We then analyze the style feature of the brush strokes and use this feature for synthesis and style transfer along the stroke curves learned from the specified area in target images or target 3D models to get the same painting style as the samples. Based on this approach, we also design and realize the corresponding painting system. The results show that the users can get a style-transferred personalized target image just by using the given sample images with the least interactions.