3D worlds are useless if the user cannot navigate and understand them. Older methods of zooming in and out, scrolling, and global bird's eye views place too much work on the user, while Fisheye Views, Perspective Wall, the Document Lens, and Desert Fog can be distorting and disorienting.
The authors' solution is a moded system which allows the user to move around in the space if they drag on empty space, or to inspect a particular object if they drag the mouse onto it. When moving around in empty space, speed is coupled with height and tilt to allow the user to seamlessly switch between local and global views.
In this system, users rarely moved backwards, and then only to go back a little bit after overshooting a target.
In a follow-up study of this idea, they tested users with monitor size as a variable. The participants were reliably faster with the 39 in. monitors than with the 15 in. monitors. Women had larger task-time differences between monitors than men did.
Navigation is 3 subtasks:
Much of this paper was not HCI oriented. The authors believe that designers should constrain user's motion to avoid the "lost-in-space" feeling that can accompany 6 degree of freedom movement, and they recommend a 3D sidewalk type metaphor that does not allow motion in any direction. They found that users retain less information about hard-to-find/see objects when they have 6 degrees of freedom than when they were using the Sidewalk metaphor. Therefore, it is important to draw attention to important objects.
Some effective techniques:
With a focus on web navigation, this paper explains how users can easily become disoriented when navigating through non-linear documents. Associative links are those within text, while contextual navigation aids are links in menus and navigation bars. User content is often categorized within either a spacial or temporal context. The best approach is to find user navigation strategies and problems, and use spacial or temporal aids to match their needs.
In study of navigation, the document's structure, size, link density, linearity, distances, and clustering should be considered. Path length, amount of backtracking, revisitation patterns and path linearity are metrics that can be used to measure user navigation characteristics.
This paper deals mainly with motion planning algorithms in 3D navigation. Navigation times were greatly decreased in their trials when an intelligent UI prevented the following troublesome phenomenon: "The controlled viewpoint gets stuck at certain locations of the scene. It can netiher move forward nore rotate at these locations without moving backward first. Users often feel frustrated with this kind of maneuvers espectialy when the frame rate is not high enough for smooth, responsive interactions."
The proposed solution to low frame rates and users running into objects and getting stuck in 3D interfaces was a many-shot randomized roadmap planner. It samples the 3D environment and makes strategies and a connectivity graph to plan possible paths of motion for the user.