In one embodiment, a technique is utilized to generate a virtual-room of a virtual room-based user interface that controls one or more devices within a physical room of a structure. A user is prompted to select one or more services provided by the one or more devices within the physical room. A plurality of images of the physical room are obtained that each share substantially a same perspective. The plurality of images include an all-off image, in which all of the one or more of the services are in an inactive state, and one or more service-specific images, in which a single service is in an active state and any other services are in the inactive state. The plurality of images are aligned to create an aligned image stack. One or more hit boxes are designated. The virtual room is generated based on the aligned image stack and the designated one or more hit boxes.