Usability analysis of 3D rotation techniques

Proceedings of the 10th annual ACM symposium on User interface software and technology - UIST '97 (1997), pp. 1-10, doi: 10.1145/263407.263408
Experiments
Tasks
Findings

How do different 3D rotation using different inputs such as the mouse-driven Virtual Sphere and Arcball, along with multidimensional inputs, affect orientation matching tasks in VE's>

Variables and Constants

    Control Symmetry
  • dimensional - half of inputs are 2 dof, the remaining are 3 dof
    System Appropriateness
  • form factor - all but one have some direct manipulation on a sphere
  • input accuracy - with the dimensionality varying and thus, constraints needed, all have varying levels of accuracy
Independent Variables
  • input device - virtual sphere, arcball, 3d ball and tracker
Constants

Display factors such as update rate (30 Hz), form factor among others, were constant

System Info

Displays
  • monitor - a 19" 1280 x 1024 true color display (0.28 mm^2 pixel size), update rate was 22 and 25 hz
Input Hardware
  • polhemus 3space 6dof tracker - sampling rate: 30 hz
  • virtual sphere - a 2d mouse with rotating a trackball. to provide the third rotational degree of freedom, a circle is drawn around the object ), and when the user clicks and drags in the area outside of the circle, rotation is constrained to be about the axis perpendicula
  • arcball - like a virutal sphere, (2d moues with trackball to directly manipulate) but with less noisy data, easy addition of constraint modes.
  • 3d magnetic ball - magnetic-tracked 3dof rotation device housed in a ball
  • magnetic orientation tracker - like 3d magnetic ball, but no housing
Software
None

An HP workstation ran the experiments

Participant Info

Mean age: 19.1 years Recruited from their psychology department's subject pool Right handed All test users had experience with the mouse, while none had any experience with 3D input devices.

Total # Age Range Gender Balance
24 -

Test users rotated the house model on the right side of the screen to match the orientation shown at the left.

Interaction and Environment

Interface

Interaction varies on input device. Please see the input device section.

The house to match was extremely simple. There were two versions of the house: the one the user had to manipulate, and the ideal orientation.

Dimensionality Scale Density Visual Realism
3D Small Low Low
Metrics

  • accuracy - accuracy was measured by the shortest-arc rotation between the final user-specified rotation and the ideal matching orientation.
  • time - completion time
  1. There was a significant interaction between dimensional and input device on time for a object manipulation task.

    3 DOF input devices were significantly faster than 2 DOF inputs for object manipulation/matching orientation tasks.

    Specificity: Somewhat general
    While the study did not explicitly test all possible 3 DOF and 2DOF input devices for rotation, the results were especially strong for the comparisons (p < .0001) The 3D ball was significantly faster than the 2D inputs (arcball & virtual sphere) and tracker was significantly faster than arcball and virtual sphere.

  2. There was a significant direct effect of input device on accuracy for a object manipulation task.

    Women were faster with a 3DOF ball than with a 2 DOF trackball device for task completion times for object orientation matching tasks.

    Specificity: Neither
    While the study did not study this explicitly, it was still a compelling result, given they did a valanced gender study with a significance of p < .05.