Fig. 5

Results from Experiment 1. (a) Participants’ preferences for each of the five projections studied (one per plot). Vote counts with a statistically significant difference (Wilcoxon signed-rank test with \(p\text{-value } < 0.05\)) are marked with an asterisk. In four of the projections, participants showed a preference for the pictures with the familiar object rendered following local linear perspective (three of them statistically significant), which seems to contradict the global consistency theory, and indeed supports the predictions from the Direct View Condition (DVC)³⁰. (b) Participants’ preference for the local linear condition w.r.t. the image-space aspect ratio of the volleyball rendered with the same projection as the scene (globally-consistent case). The more distorted the sphere (i.e., the more different the projection was w.r.t. linear perspective), the less probable it was for participants to choose it (Pearson’s \(r = 0.95\), Spearman’s \(\rho = 0.89\)); this seems to suggest that people tend to prefer those projections where the object is closest to its normal projection, again contradicting global consistency and supporting DVC.