Where front projection is concerned the image, as seen by the viewer, is projected onto a reflecting screen. The viewer sees the image reflected from a specifically coated screen.
By the unfavourable selection of the location of the viewer(s) overlapping might be experienced. As a rule this overlapping is avoided by mounting the projection unit to the ceiling and/or placing the viewers behind or next to the projectors. Thus, for instance, a counter can serve as separation between the projection room and the viewer and the projection technology is hidden. Such a set-up is very suitable for fairs for instance.
In the case of rear projection, the image is displayed on the rear of the projection screen (seen from the viewer). Hereby overlapping or shadows of the viewer or other persons are avoided. Rear projection units are therefore better suited for interactive scenarios. As the user in fact looks into the light ray, the same projectors generally produce more brilliant images than they do in the case of front projection.
The arrangement of the projection behind the screen calls for additional space. This space requirement can be minimised by deflecting the optical path with suitable mirror constructions consisting of one or several mirrors.
If several projection surfaces are combined into one environment, you get a multi-plane projection. By the suitable arrangement of the individual sides you get 3D environments which extend the user's viewing field. So-called caves or panorama walls are possible means of application. The challenge is the synchronous actuation of all projection surfaces.
When distributing the installations the interactions and manipulations on a model can also be transmitted to the remote location via narrow band communication channels. Hereby, for instance, a spatially distributed design review can be put into effect.
The generation of depth impression is effected in the human brain, among others with images that are different for the right eye and for the left eye (parallax). This has to be copied in the computer graphics; to all intents and purposes, 2 separate images have to be shown on one projection surface. Both the images have to be generated for the respective eye and be displayed in such a manner that only the eye concerned sees the image that it is supposed to see. To achieve this end there are two principal means:
- Theseparate images are displayed in sequence with a frequency of 100 Hz to 120 Hz. The light path for the left eye is scanned dark by a shutter when the "right image" is shown on the screen - and vice versa. This technique is defined as active channel separation or shutter technique.
- Both images are projected permanently. The separation of the channels for the left and the right eye is passive (i.e. without switchover) for example by polarisation filter or colour filter (red-blue filter) or so-called comb filters.
Passive channel separation makes it possible to use standard projectors at a favourable price. The development of this technique and the corresponding software for the graphics control was the basic prerequisite for the implementation by small businesses and will make its fundamental contribution to the spreading of VR as practicable tool.
When using linear polarising filters for passive channel separation exactly such oscillation level of the light is filtered that concurs with the light of the polarising filter from the non-polarised light entering.
If linear polarised light meets a polarising filter, the orientation of the filter decides whether the light may pass or is blocked out. In the case of parallel orientation the light can shine through unimpaired, in the case of vertical orientation the light is blocked. If two filters are arranged vertically next to each other, the optimal separation of the channels from each other is achieved.
Non polarised light is oriented in the form of a screw length in circular polarisation. The channel separation is achieved by the varying orientation of the screw length; one filter turning clockwise and one anti-clockwise are combined.
Contrary to the linear filters, the change of head inclination has no effect on the channel separation (no 'ghost images' when head is tilted or position changed). Hence these filters are suitable for users who interact actively with the model
One the other hand the channel separation is not as good as for a linear polarisation.
A correct display of the depth impression in a stereoscopic presentation calls for info about the location or viewing path of the user. Then, based on the eye spacing and the distance of the viewer to the screen, the images for the left eye and the right eye are calculated. In the simplest case, a standard viewing position is assumed from which the optimal spatial depth effect is visible.
In an interactive application in which the model can also be seen from the sides or underneath, the location of the user has to be updated continually, in line with the user's movements. The data is then automatically acquired; the user is 'tracked'.
Tracking can be directly integrated in the VR software. However, the variant of a separate tracking server which operates independently and the data of which, for instance, is called up via the local network and is integrated into the calculations of VR output is more flexible. The precision and action radius of the position determinations can be considerably improved by the use of several cameras.