Fig 15. Relationship of mean eye-tooth distance and learning gains.

(a) All samples show a weak correlation. (b) Influence of 3D rendering mode. Both correlations are weak. (c) Influence of hand-tool alignment. Alignment shows a moderate positive correlation. (d) Influence of condition groups. Group 2 (alignment & mono 3D) shows a strong positive correlation and group 4 (misalignment & stereo 3D) a strong negative correlation. The samples in the mono 3D have a global minimum and maximum at either extremes, whereas stereo 3D has a global minimum in the middle and the performance to the extremes gets worse. This suggests that there is an optimal distance for stereo 3D, a value after which the stereo vision suffers because of the large stereo disparity. For mono 3D (& aligned) the shorter the distance to the tooth, the better the learning performance.