IPTs allow small group interaction within the space
of the IPT. The most common demonstration scenario, as shown in Figure 2b,
is one “pilot” head-tracked with another demonstrator in control of
the situation. In this situation the demonstrator may need to indicate
items of the display or control the environment, but they are seeing a
distorted view, as shown in Figure 3.
An experiment was undertaken in May 2003 to compare
head-tracked verses non-head tracked tasks performance [22
]. It had been suggested by others that interaction on IPTs for
non-head tracked users would be either difficult or impossible. We showed
that actually the non head-tracked user (typically the demonstrator) could
interact very rapidly with a suitable choice of interaction metaphor.
We introduced a new interaction metaphor called
shadow cone selection. All the objects with a cone are selected when the
user presses a button, and as the user moves their hand only objects that
are always within the cone are selected when the button is released. To
implement this, when the user presses the button, the potential selection
set is initialised with all objects that are within a target angle of the
ray from the hand. On each frame we remove any object is no longer within
the target angle and we don’t add new objects. When the user releases
the button the objects remaining in the potential selection set are then
selected. This can be visualized by imaging turning on a spotlight in a
dim scene and selecting only the objects that are always in the spotlight
before it is turned off.
The experiment compared shadow cone, standard cone and ray casting on
selection tasks. The results were complicated because no technique stood
out as the best on all tasks. The results can perhaps best be conveyed by
the guidelines we proposed in the associated paper:
·
Virtual hand should only be used for SIDs when the user
will be head-tracked. We suggest that implementers avoid it for SIDs
because of the variety of viewing scenarios.
·
Ray, Cone and Shadow Cone selection are all suitable for
non-head tracked SID users. Go-Go hand, image plane selection or aperture
selection are not.
·
Close-by manipulation should only be attempted with
head-tracked egocentric views. For the non-head tracked view, only
manipulation at a distance should be attempted.
We can also propose
some tentative guidelines:
·
Consider using Cone Selection or Shadow Cone Selection in
preference to Ray Selection if there is any significant jitter in the
tracking system.
·
Shadow Cone Selection and Ray Selection are more suitable
than Cone selection for complex tasks involving clustered objects.