We developed a prototype which overlays local and remote participants in a video call and enables them to take group pictures together. These pictures serve as keepsakes of the event. The application uses real-time chroma key background removal to composite the remote person into the scene with the local group. We tested the prototype in a museum setting, and compared it to a more standard picture-in-picture (PiP) model. Users rated the composite mode as being significantly more fun, creating a greater sense of copresence and involvement than the PiP mode. Composite snapshots were also strongly preferred over picture-in-picture. Based on results from the study, we added a pinch-zoom and positioning interface to make it easier to frame remote people together into the snapshot, and conducted a second study. We conclude that combining composite video calls and picture-taking on a mobile device enables people to socially construct a shared activity with a remote person.