Abstract: We introduce two types of image representations for modeling the composition styles and the tones of photos. Based on the compositional image representation, we present an algorithm that imitates the composition styles of 'National-Geographic' landscape photography and finds good views within a panoramic scene for taking pictures. Given an initial location in a panoramic scene, our algorithm is able to suggest a better view that would often yield professional-like photo composition. For color and tone management, we present a sparse color representation of images. This sparse color representation allows us to do color transfer and color correction on images more effectively. Many examples and experimental results will be shown in the talk to illustrate the ideas and the advantages of our approaches.