Game Development Reference
Improvements to the subdivision algorithm and resulting data amplification
could be made to better utilize hardware depth culling when rendering the re-
sulting mesh. Increasing the subdivision granularity per iteration, and adding
the subdivided regions to the output streams in increasing order of average depth
(nearest to furthest) would enable the final rendering pass to benefit more from
Combining this technique with the built-in hardware tessellation now avail-
able in modern graphics hardware could have some benefits, perhaps resulting
in increased frame rates. Once a minimally acceptable level of subdivision has
been achieved through our iterative process described in Stage 2 of Section 1.2.2,
a single additional pass utilizing hardware tessellation may reduce unnecessary
computations. Further research into this possibility would be needed to measure
[Ebert et al. 98] D. S. Ebert, F. K. Musgrave, D. Peachey, K. Perlin, and S.
Wo r ley. Texturing and Modeling: A Procedural Approach , Second Edition.
San Diego: Academic Press, 1998.
[Strugar 10] Filip Strugar. “Continuous Distance-Dependent Level of De-
tail for Rendering Heightmaps (CDLOD).” http://www.vertexasylum.com/
downloads/cdlod/cdlod latest.pdf , 2010.
[Johanson 04] Claes Johanson. “Real-Time Water Rendering Introducing the
Projected Grid Concept.” Master of Science Thesis, Lund University, 2004.
(Available at http://fileadmin.cs.lth.se/graphics/theses/projects/projgrid/