Proceedings: GI 2004

Compressed multisampling for efficient hardware edge antialiasing

Philippe Beaudoin , Pierre Poulin

Proceedings of Graphics Interface 2004: London, Ontario, Canada, 17 - 19 May 2004, 169-176

DOI 10.20380/GI2004.21

  • Bibtex

    author = {Beaudoin, Philippe and Poulin, Pierre},
    title = {Compressed multisampling for efficient hardware edge antialiasing},
    booktitle = {Proceedings of Graphics Interface 2004},
    series = {GI 2004},
    year = {2004},
    issn = {0-89791-213-6},
    isbn = {1-56881-227-2},
    location = {London, Ontario, Canada},
    pages = {169--176},
    numpages = {8},
    doi = {10.20380/GI2004.21},
    publisher = {Canadian Human-Computer Communications Society},
    address = {School of Computer Science, University of Waterloo, Waterloo, Ontario, Canada},
    keywords = {Hardware},


Today's hardware graphics accelerators incorporate techniques to antialias edges and minimize geometry-related sampling artifacts. Two such techniques, brute force supersampling and multisampling, increase the sampling rate by rasterizing the triangles in a larger antialiasing buffer that is then filtered down to the size of the framebuffer. The sampling rate is proportional to the number of subsamples in the antialiasing buffer and, when no compression is used, to the memory it occupies. In turn, a larger antialiasing buffer implies an increase in bandwidth, one of the limiting resources for today's applications. In this paper we propose a mechanism to compress the antialiasing buffer and limit the bandwidth requirements while maintaining higher sampling rates. The usual framebuffer-related functions of OpenGL are supported: alpha blending, stenciling, color operations, and color masking. The technique is scalable, allowing for user-specified maximal and minimal sampling rates. The compression scheme includes a mechanism to nicely degrade the quality when too much information would be required. A lower bound on the quality of the resulting image is also available since the sampling rate will never be less than the user-specified minimal rate. The compression scheme is simple enough to be incorporated into standard hardware graphics accelerators. Software simulations show that, for a given bandwidth, our technique offers improved visual results over multisampling schemes.