Mixed-initiative photo collage authoring

Creating an artifact that captures the story or memory from a large photo collection is a difficult task, because the tools available are either too difficult to learn, or oversimplified to the point that they lack flexibility. Individual techniques have been developed to automate parts of the selection-editing-composition cycle, but relatively little has been done to strike the right overall balance between the fully automatic and the fully manual. In this paper, we present miCollage, which attempts to piece together individual technologies to create a compelling collage authoring experience. The system consists of three main components. In the selection component, the system can make proactive suggestions about which photos to add to the collage as well as help the user to find similar or related photos. In the editing component, the system applies automatic cropping and enhancement to the images. In the layout component, the system suggests alternative layouts but is still able to accommodate manual changes, while satisfying various spatial constraints. The user interface connects the components seamlessly, allowing a best of both worlds between fully manual and fully automatic collage authoring.