We present a system that enables a novice user to acquire a large indoor scene in minutes as a collection of images sufficient for five degrees-of-freedom virtual navigation by image morphing. The user walks through the scene wearing an augmented reality head-mounted display (AR HMD) enhanced with a panoramic video camera. The AR HMD shows a 2D grid of a dynamically generated floor plan, which guides the user to acquire a panorama from each grid cell. After acquisition, panoramas are preliminarily registered using the AR HMD tracking data, corresponding features are detected in pairs of neighboring panoramas, and the correspondences are used to refine panorama registration. The registered panoramas and their correspondences support rendering the scene interactively with any view direction and from any viewpoint on the acquisition plane. An HMD VR interface guides the user who optimizes visualization fidelity interactively, by aligning the viewpoint with one of the hundreds of acquisition locations evenly sampling the floor plane.