![](bg6.png)
core rendering mechanisms. The footprint of the IBA
can be considered constant. It depends on the tex-
ture resolution s, the precision per color channel b,
the number of color channels c, and the number of
raster layers l. The footprint can be approximated
by: O
IBA
(l, s, b, c) = 2 · l · c · b · s
2
byte without mip-
maps. This parametrization enables the user to bal-
ance the trade-off between image quality and memory
consumptions, as well as the runtime complexity.
The memory footprint of the GBA is dynamically
view-depended and scales linearly with the number
of input triangles t. Further, memory footprint de-
pends on the average rate of primitive amplification r
(for a 180
◦
cylindrical projection r = 1.5 −2), and the
size of each triangle in an intermediate data structure
i = 16 byte. The amount of additional memory can
be approximated by: O
GBA
(t, r, i) = t · r · i. Following
to that, the space complexity of the GBA is indepen-
dent of rendering a single NPP or a stereo pair of NPP.
For the complex model (3,210,162 triangles) the addi-
tional memory requirement for a 180
◦
panorama pro-
jection is O
GBA
=∼ 69 MB. This corresponds to four
RGBA raster layers with 1024
2
pixels resolution. For
a higher FOV: O
GBA
< O
IBA
is valid in any case.
6 CONCLUSIONS
This paper presents an overview for creating stereo
renderings of non-planar projections with image-
based and geometry-based rendering techniques. In
particular, it describes the implementation of a single-
pass image-base rendering technique as an extension
to an existing framework. We evaluate the perfor-
mance of this technique with respect to the number of
input triangles. We further present a comparison be-
tween geometry-based and image-based approaches
for generating stereo pairs with respect to of four dif-
ferent criteria.
This comparison shows that both approaches are
capable of rendering stereographic non-planar projec-
tions. The GBA is predominant over IBA in the range
of functionality with respects to stereo rendering, the
quality of the output images, as well as the render-
ing performance. The IBA has advantages consider-
ing the constant space and low implementation com-
plexity. A disadvantage of both rendering techniques
is the limitation to polygonal scenes only. They can-
not be applied directly to volume rendering without
major changes.
Following to these results, we consider the GBA
more suitable for stereo rendering of non-planar pro-
jections than the IBA. According to our performance
measurements, both approaches achieve satisfying
results for 3D scenes of the medium complexity
(500,000 triangles).
REFERENCES
Bimber, O., Wetzstein, G., Emmerling, A., and Nitschke, C.
(2005). Enabling View-Dependent Stereoscopic Pro-
jection in Real Environments. In IEEE ISMAR ’05,
pages 14–23, Washington, DC, USA. IEEE Computer
Society.
Blythe, D. (2006). The Direct3D 10 System. In SIGGRAPH
’06: ACM SIGGRAPH 2006 Papers, pages 724–734,
New York, NY, USA. ACM Press.
Boubekeur, T. and Schlick, C. (2008). A Flexible Kernel
for Adaptive Mesh Refinement on GPU. Computer
Graphics Forum, 27(1):102–114.
Bourke, P. D. and Morse, P. (2007). Stereoscopy, Theory
and Practice. Workshop, VSMM 2007, Brisbane.
Hou, X., Wei, L.-Y., Shum, H.-Y., and Guo, B. (2006).
Real-time Multi-Perspective Rendering on Graphics
Hardware. In EUROGRAPHICS Symposium on Ren-
dering. Blackwell Publishing.
Kessenich, J. (2006). The OpenGL Shading Language Lan-
guage Version: 1.20 Document Revision: 8.
Lorenz, H. and D
¨
ollner, J. (2008). Dynamic Mesh Refine-
ment on GPU using Geometry Shaders. In Proceed-
ings of the 16-th WCSG 2008.
NVIDIA (2008). NVIDIA OpenGL Extension Specifications
for the GeForce 8 Series Architecture (G8x). NVIDIA.
Peleg, S., Ben-Ezra, M., and Pritch, Y. (2001). Omnis-
tereo: Panoramic Stereo Imaging. IEEE Transac-
tions on Pattern Analysis and Machine Intelligence,
23(3):279–290.
Persson, E. (2007). ATI Radeon HD 2000 Programming
Guide. AMD Graphics Products Group.
Spindler, M., Bubke, M., Germer, T., and Strothotte, T.
(2006). Camera Textures. In Proceedings of the 4th
GRAPHITE, pages 295–302. ACM.
Tatarchuk, N. (2007). Real-Time Tessellation on GPU.
In Course 28: Advanced Real-Time Rendering in 3D
Graphics and Games. ACM SIGGRAPH 2007.
Tatarinov, A. (2008). Instanced Tessellation in DirectX10.
In GDC ’08: Game Developers’ Conference 2008.
Trapp, M. and D
¨
ollner, J. (2008). A Generalization
Approach for 3D Viewing Deformations of Single-
Center Projections. In Jos
´
e Braz, N. J. N. and Pereira,
J. M., editors, GRAPP 2008, number 3, pages 162–
170. INSTICC Press.
Yang, Y., Chen, J. X., and Beheshti, M. (2005). Nonlinear
Perspective Projections and Magic Lenses: 3D View
Deformation. IEEE Computer Graphics and Applica-
tions, pages 76–84.
GRAPP 2009 - International Conference on Computer Graphics Theory and Applications
204