Abstract: When a 3D model is transmitted over a lossy network,some model information may inevitably be missing. Under such situation,one may not be able to visualize the receiving model unless the lostmodel information has been retransmitted. Progressive modeltransmission offers an alternative to avoid the ``all or nothingsituation'' by allowing a model to be visualized with a degraded qualitywhen only part of the model data has been received. Unfortunately, incase some model refinement information is missing, one may stillneed to wait for such information to be retransmitted before the modelcan be rendered with a desired visual quality. To address this problem,we have developed a novel error resilient packetization scheme. Wefirst construct a Non-Redundant Directed Acyclic Graph to encode thedependencies among the vertex splits of a progressive mesh. A specialGlobal Graph Equipartition Packing Algorithm is then applied topartitioning this graph into several equal size sub-graphs, which ispacked as packets. The packing algorithm comprises two main phases:initial partition phase and global refinement phase. Experimentalresults demonstrate that the proposed scheme can minimize thedependencies between packets. Hence, it reduces the delay in rendering3D models with proper quality at the clients.