Recent advances in image-based rendering techniques allow us to have interactive panoramic viewing of real world environment using real photos. With panorama visualization tools like QuickTime VR or Real-space Viewer, we are able to have fast navigation in an image-based environment. However, the navigation control is restricted to panning, zooming and tilting of a single panoramic view. Smooth walkthrough between panoramic nodes is not yet supported.In this paper, we first describe how we apply the epipolar model between cylindrical projection manifolds to find pixel correspondence between real-world panoramic photos. And a correspondence matching program is written to allow the users to identify pixel correspondence interactively. Then, this program will construct one triangular mesh for each neighboring panorama and the users can then identify layered patches on it. After that, we are able to provide free navigation between panoramic nodes by means of the triangle-based warping technique and the result is shown to be occlusion-compatible.