Effect rendering that renders 3D model to 2D images with various coloring and lighting effects, is an important step in home interior design. Traditional way of manual rendering using professional software is very labor intensive and time consuming. In this paper, we present a novel capsule based conditional generative adversarial network that can automatically synthesize an indoor image with realistic and aesthetically pleasing rendering effect from a given plain image rendered without any effects from a interior designed 3D model. By adapting capsule blocks in both generator and discriminator and a novel multi-way loss function inside discriminator, our framework is able to generate more realistic rendering effect at both detail and global levels. In addition, a novel line preservation loss is introduced not only to help preserve the properties that are independent of lighting effect, but also improves the lighting effect along those lines. We apply our technique on a dataset specially prepared for interior design effect rendering and systematically compare our approach with multiple state-of-the-art methods.