The Spatio-temporal coherence of time-varying model sequence is an important and highly-desirable property in varies of applications. We propose an efficient framework that reconstructs a Spatio-temporally coherent point cloud sequence for dynamic objects from multiview video. The main distinction of the framework is that it does not complicatedly and strictly match between any two contours on the same model or the neighboring model. The point cloud between the discrete frames is obtained by transporting the neighboring models or by reconstructing the in-between model based on interpolation silhouette. The time-varying and quasi-dense point cloud of such dynamic objects shape can be tracked successfully thanks to Spatio-Temporal-Contour consistency and distance field interpolation employed in our shape tracking framework. Compared to existing approaches, the resulting point cloud sequence has better Spatio-temporal coherence. Experimental results demonstrate that the framework obtains the promise.