3D Mesh Streaming

3D triangular mesh is becoming an increasingly important data type for networked applications such as digital museums, online games, and virtual worlds. A high quality 3D mesh can consists of billions of polygons, giving a total data size in the order of gigabytes, even after lossless compression. Disseminating such 3D meshes over a resource-constraint network for on-demand viewing and rendering presents many technical challenges.

This project is partially supported by MOE ARF Grant 232-000-306-112. (September 2007 - September 2009)

  • Wei Tsang Ooi
  • Wei Cheng
  • Dan Liu
  • Sebastien Mondet (IRIT, University of Toulouse)
  • Romulus Grigoras (IRIT, University of Toulouse)
  • Geraldine Morin (IRIT, University of Toulouse)
  • Jeremie Biancheri (INSA Leon, Summer 2008)
  • Yan Gu (NUS, 2004-2005)
  • Tran Hong Quang (NUS, 2004)
  • Receiver-Driven View-Dependent Streaming of Progressive Mesh Wei Cheng, Wei Tsang Ooi, In Proceedings of the 18th International Workshop on Network and Operating System Support for Digital Audio and Video (NOSSDAV), 28-30 May 2008. 17 out of 50 (34%) [pdf] [slides]
    Existing mesh streaming systems use a sender-driven approach, where the receiver informs the sender of its viewpoint, and the sender checks for vertex visibility and sends the visible vertices in decreasing order of visual contributions. This approach places high computational load on the sender, limiting its scalability. Furthermore, the sender needs to maintain states for each receiver, remember what each receiver has received. We propose a receiver-driven approach in this paper, where the receiver performs visibility checks and explicitly request vertices from the sender. The sender is stateless and merely returns the requested vertices. The challenge addressed in this paper is (i) how to efficiently encode the request for vertices, and (ii) how the receiver can estimate the visual contributions of new vertices to request based on partially reconstructed mesh.
  • An Analytical Model for Progessive Mesh Streaming Wei Cheng, Wei Tsang Ooi, Sebastian Mondet, Romulus Grigoras, Geraldine Morin, To appear in
    Proceedings of 15th ACM International Multimedia Conference, September 24-29, 2007, Augsburg, Germany. [pdf] [slides]
    In this paper, we present an analytical model to estimate the quality of the received mesh as a function of dependencies among the mesh and network parameters. The effect of dependency in 3D models is different with the one in video streaming, so retransmission is important. With retransmission imediately after we detect the packet loss, the effect of dependency still exist, but it only matters in the first few seconds. We can further reduce the dependencies among packets by using a heuristic based on our analytical model to packetize the mesh.
    The following two videos demonstrate the visual difference in results when the same mesh is transmitted using the naive FIFO scheme (left) and our heuristic (right).
  • Packetization of 3D Progressive Meshes for Streaming over Lossy Networks Yan Gu, Wei Tsang Ooi, ICCCN 2005, San Diego, CA, 17-19 October 2005. Best Paper Candidate [pdf]
    This paper is the first that poses the packetization problem for progressive mesh and show that it can affect the received mesh quality. We propose a greedy algorithm that aims to reduce the dependency among packets. Importance of vertices and network conditions are not considered in this paper (see our ACM MM 2007 paper for an improvement).
  • Dependency-aware Packetization Problem in Progressive Meshes, Yan Gu, INFOCOM 2005 (Poster), Miami, FL, March 2005.
    This short paper shows that minimizing the dependencies among packets when packetizing a progressive mesh is NP-complete.