Supplemental Material forreference and Artifact Analysis for Video Transitions of Places

These transitions are strongly disliked under both slight and considerable view changes, and our experimental findings suggests that cuts are not appropriate for our system. This is expected as cuts provide no additional cues to aid the orientation of the viewer. From the 30◦ rule, we might expect cuts to be preferred more in considerable view changes than slight view changes. However, for these scenes our data suggests otherwise: the cut transition was unpreferred against fewer transitions in the slight case (with plane and APC having insignificant differences, Table 2b), and the absolute difference in preference between the nearest transitions is greater in the considerable case (Figure 1, 0.16 for slight vs. 0.80 for considerable). Further observation reveals that in the plane and APC cases, we can see that this difference in significance comes from the variance due to artefacts which appear more objectionable in certain scenes, specifically sets 5 and 6 for the plane transition (skewed scenes and large empty areas) and set 7 for the APC transition (pepper noise).