In this paper, we explore the more practical aspects of building and rendering concentric mosaics. First, we use images captured with only approximately circular camera trajectories. The image sequence capture can be achieved by holding a camcorder in position and rotating the body all around. In addition, we investigate the use of variable input sampling and fidelity of scene geometry based on the level of interest (and hence quality of view synthesized) on the objects in the scene. We achieve the tolerance for minor perturbations about the exact circular camera path and variable input sampling by using and analyzing a variant of the Hough space of all captured rays. Examples using real scenes are shown to validate our approach.