The proposed algorithm attempts to combine the partitioning based 3D dynamic data compression and the multi-level k-way graph partitioning algorithms to achieve multi-resolution reconstruction of animation sequences. Partitioning based 3D dynamic data compression efficiently encodes changes between partitioned meshes while the multi-level k-way graph partitioning algorithm provides partitioning information for the input graph (mesh) at various resolutions. Combining the two provides a powerful tool to incorporate both spatial down sampling and temporal compression in a streaming framework where the server can satisfy the needs of a low-end user as well as effectively adapt to dynamically varying bandwidth constraints. At the decoder, the mesh may either be reconstructed at the streamed resolution or at a higher resolution by ingeniously adding vertices so that the viewer does not feel the effects of down sampling, and consequently, the deterioration in quality. Our algorithm deals with meshes whose connectivity does not change with time.