On page 4-5 in "
Modern Approaches to Augmented Reality", a course at SIGGRAPH 2005, Bimber and Raskar writes:
"
Head-mounted projective displays (HMPDs) decrease the effect of inconsistency of accommodation and convergence that is related to HMDs. Both, head-mounted projective displays and projective head-mounted displays (PHMDs) also address other problems that are related to HMDs: They provide a larger field of view without the application of additional lenses that introduce distorting arbitrations. They also prevent incorrect parallax distortions caused by IPD (inter-pupil distance) mismatch that occurs if HMDs are worn incorrectly (e.g., if they slip slightly from their designed position). However, they also introduce several
shortcomings:
• Both, head-mounted projective displays and projective head-mounted displays are currently cumbersome. However, new prototypes tend to be smaller and more ergonomically to wear;
• The integrated miniature projectors/LCDs offer limited resolution and brightness;
• Head-mounted projective displays might require special display surfaces (i.e., retro-reflective surfaces) to provide bright images;
• For projective head-mounted displays, the brightness of the images depends on the environmental light conditions;
• Projective head-mounted displays can only be used indoors, since they require the presence of a ceiling.
Although such displays technically tend to combine the advantages of projection displays with the advantages of traditional HMDs, their cumbersomeness currently prevents them from being applicable outside research laboratories. Like head-attached displays in general, they
suffer from the imbalanced ratio between heavy optics (or projectors) that results in cumbersome and uncomfortable devices or ergonomic devices with a relatively poor image
quality.
"
Several of the shortcomings have diminished since the review was written about two years ago: Projectors have become/are becoming much smaller and (green) laser diodes are becoming generally available. With laser diode driven projectors the brightness/energy needed should be much better.
HMPDs still require special (retro-reflective) material and in my point of view this is what makes them great and not one of their shortcomings as stated above. Retro-reflective material makes it possible to create independent views for each viewer, independently of the direction they are looking at the material. This means that several users can stand anywhere around a display surface and still see independent views. This is not possible with the
tabletop kind of display technology I have worked with the last couple of years, where each independent view is limited to one edge of the
square tabletop.
One shortcoming that was not mentioned in the HMPD review is that of tracking. The same shortcoming as stated for optical see-through AR should also apply for HMPD's:
"Optical see-through devices require difficult (user and session dependent) calibration and precise head-tracking to ensure a correct graphical overlay.", see page 4 in "
Modern Approaches to Augmented Reality".
One example of solving the tracking is presented in
Hua et al. paper describing the HMPD environment SCAPE, there they use a
HiBall 3000, which is a pretty fast (up to 2000hz) and precise (0.2 mm positional res). The HiBall tracking speed and resolution is great. In fact I do not know of any 6DOF tracker coming close to it's performance. One problem is that it not very portable. You have to bring your roof mounted HiBall Beacon Array Modules (and a roof) and stick a HiBall sensor on top of your head.
To not only look HMPD cool...
...but also have a useful tool for Augmented Reality, high precision tracking is a requirement. But how do you create a tracker that has the performance of the HiBall and still can be hidden in a pair of cool eyewear to go anywhere you go?
Now you reached the speculation part of this post:
The answer to the above question is that you don't! At least you do not need to track your head's position and the direction you face using an external tracker. With display technology such as the MVIS PicoP capable of scanning a laser beam across objects in the environment 60 times a second with high precision it seems obvious to me that you instead keep track of the distance and direction to the objects in relation to the laser beam. The display becomes the tracker. To triangulate the position of the laser beam you also need at least one image sensor.
Theory and
Practice.
Labels: co-located, doktorand, hmpd, phd, view-dependent, visualization
