Graphics Reference
In-Depth Information
60
Without impostor
55
30ms (1 MB)
50
25ms (2 MB)
45
20ms (6 MB)
40
15ms (18 MB)
35
10ms (103 MB)
30
Env. maps (233 MB)
25
20
15
10
5
0
2401
1
401
1201
1601
2001
801
Frame
FIGURE 8.3
Experiment results from Jeschke et al.'s approach [77] with usage of imposters.
rate transitions because it can severely affect the user experience due to the “display
stuttering” that occurs. With reference to the diagrams in Figure 8.7, Zheng et al.
[76] showed in their experiment results that their algorithm in handling distributed
rendering produces frame rate accuracies close to the targets. This is however done
with bumpy transitions and as depicted in the diagram on the left, there are even
oscillations after an initial steady-state.
In contrast to the results shown in Figure 8.7, our fuzzy controller system produces
tracking with improved transitions as shown in Figures 5.12, 5.13 (now Figures 8.5
and 8.6) and 6.11 in Chapter 6. There are no sporadic oscillations with large ampli-
tude after the output attains a steady-state level.
8.1.3 a daPtiVe t Racking c aPaBility
While research in interactive 3D rendering purports accurate tracking to a perfor-
mance objective, what is often not presented is the ability of the technique to adapt to
changing performance objectives. We illustrate a practical example where an appli-
cation that may draw considerable computer hardware processing power can benefit
from a longer run-time if the display frame rate can be adaptively changed according
 
Search WWH ::




Custom Search