Game Development Reference
In-Depth Information
Image Space
The effective image space algorithms are becoming a more and more important
way of achieving more realism or higher quality of the final image.
We start the section with two articles improving the depth-of-field effect used
in many modern games. The first article is “The Skylanders SWAP Force Depth-
of-Field Shader,” by Michael Bukowski, Padraic Hennessy, Brian Osman, and
Morgan McGuire. It describes the depth-of-field shader used in production at
Vicarious Visions for the Skylanders series of games. Their technique generates
very convincing near and far out of focus areas completely in image space without
any additional scene rendering.
The second article “Simulating Partial Occlusion in Post-Processing Depth-of-
Field Methods,” by David C. Schedl and Michael Wimmer uses the ideas similar
to order independent transparency methods to store multiple depth layers of
the rendered scene. Having the multiple depth layers allows for more realistic
rendering of the out of focus areas of the scene.
“Second-Depth Antialiasing,” by Emil Persson discusses novel semi-analytical
antialiasing method that uses the regular depth buffer and a new second-depth
depth buffer to precisely identify the geometry edges and the amount of an-
tialiasing they need. The author provides detailed implementation information,
performance analysis and full source code on the accompanying DVD.
The next article is “Practical Frame Buffer Compression,” by Pavlos Mavridis
and Georgios Papaioannou. Authors describe a lossy buffer compression method
based on the principles chrominance subsampling. The method provides a prac-
tical way of reducing bandwidth and improving associated performance, which
are required in the modern high-resolution games. It allows for direct rendering
into two channel render targets including alpha blending. The authors discuss
multiple methods of reconstruction of the regular RGB data.
The last article, “Coherence-Enhancing Filtering on the GPU,” by Jan Eric
Kyprianidis and Henry Kang, shows CUDA implementation of a fully automatic
image filter, which aggressively smoothes out the less important image regions
while preserving the important features. The authors provide extensive back-
ground for the filtering along with very detailed implementation guidelines.
I would like to thank all authors for the effort they put into their articles and
for the novel, inspiring ideas that go beyond regular polygon rendering.
—Michal Valient
Search Nedrilad ::

Custom Search