overview
design philosophy
 
Projects People Publications software Contact
 
 
  design philosophy

 
  The Visual Interaction Platform (VIP) is a working prototype incorporating some new interaction techniques for the navigation through and the manipulation of both 3D and 2D data. The prototype aims at professional applications like architectural design, surgical planning and geological exploration. Its design was influenced by the analysis of user requirements and by the requirement for a natural interface.


In order to incorporate our idea of naturalness, we adopt the following five design guidelines for interaction devices:
Guideline
Description
DG1
Two-handed interaction
Two-handed interactions should be preferred over one- handed interaction. The users know where their hands are relative to one another and this can help to develop interactions which are less demanding of visual attention.
DG2
Visual feedback
Visual feedback is important for creating a feeling of presence or spatial awareness.
DG3
Action and Perception space
The action and the perception space should coincide.
DG4
Minimal use of intrusive devices
Minimal use of intrusive devices, such as head-mounted displays, should be preferred.
DG5
Wireless props
Wireless props are preferred over wired ones. A prop is a physical device, which allows the user to interact with the data. An interface includes one or more props.














In an effort to get a better grasp on the navigation and manipulation tasks of interest to us, we have adopted the following operational definitions:
Definition
Description
2D Navigation
This involves finding items in a 2D dataset (such as a collection of pictures). Finding is typically accomplished by positioning the viewpoint sufficiently closer to the item such that the resolution and the field of view (FOV) of the displayed image are such that identification becomes possible. 2D navigation is hence required in instances where not all items can be rendered simultaneously with the necessary resolution. Depending on whether the items are arranged in a 1D or 2D array, positioning requires either 1 or 2 DOF.
3D Navigation
 This is in principle very similar to 2D navigation. Practical problems arise in current interfaces because the positioning of the viewpoint has to be done in 3D, and because more parameters are involved in creating the displayed image. Especially in case of scientific data, such as the medical CT data or the geological seismic data mentioned before, deciding on an adequate display mechanism is a problem in itself. One possible approach is to map the scientific data to a surface model. The required segmentation not only reduces the available information, but may also be incorrect and lead to a biased or erroneous interpretation. The alternative that we propose here is to navigate by means of dynamically updated cross-sections. This implies that time makes up for the missing dimension. We prefer this approach because it preserves all information in the data. A rigid rectangular frame that can be positioned freely in space will be used to create such intersections. This frame will then have 6 DOF.
2D Manipulation
 This task demands the atomic actions of positioning (3 DOF: 2 translations and 1 rotation), selecting and sketching/writing in 2D. It is familiar from existing desktop interfaces.
3D Manipulation
 Users can perform preci-sion manipulation tasks better in 2D than in 3D space. Based on this assumption it may be worthwhile to consider performing 3D manipula-tions through a series of 2D manipulations. For example, a path in 3D could be created by letting the computer interconnect user-specified points in a number of intersection images.