The out-the-window view in commercial flight simulators is a reflection seen in a large curved mirror which surrounds the simulator cabin. The projection screen displaying the imagery is on top of the cabin out of direct view. This seemingly ungainly configuration is called a cross-cockpit collimated display. Its use is mandated by civil aviation authorities because collimation provides the same perspective view regardless of the viewing location in the cabin. The intent is that all crew members receive the same out-the-window visual clues of aircraft attitude.
Collimation does this by causing all light ray paths from a given screen pixel to be parallel. No matter where the viewer is in the cabin, the light from that pixel appears to be coming from the same direction. The light diverges widely as it initially leaves the screen, but being reflected by the curved mirror changes that. The collimation magic is in the mirror shape.
The majority of commercial collimated displays use spherically curved mirrors. A spherical surface does not collimate exactly. We normally think of parabolic mirrors as perfect collimators; however, if the spherical curve is not too deep, collimation is very good. A spherical shape has the advantage of being more symmetric and not forcing a preferred viewing location when used in a collimation system. Further, large spherical mirrors are relatively easy to make compared to other shapes.
Collimation mirrors have been made from metal, glass, and rigid plastic, but contemporary collimating mirrors are generally made of reflective Mylar film. A carefully shaped framework supports the edge of the film over a closed cavity. Ambient air pressure forces the film into the cavity as a partial vacuum is pulled in the cavity.
Think of a soap bubble. The uniform surface tension of the soap film naturally pulls the bubble into a sphere. Elastic films, both liquid and solid, tend to distribute tension uniformly when stretched. As long as it isnít stretched too far, Mylar film behaves the same way, and will naturally form a spherical shape in response to differential air pressure.
(Much of the most recent work on developing collimated displays with very large vertical fields of view centered on issues which resulted from Mylar being stretched beyond its elastic limits.)
The use of a spherical collimating mirror requires the use of a nearly spherical-section projection screen having a radius of somewhat more than half that of the mirror. The shape and position of the screen has a large impact on the collimation accuracy of the display system, so the specific dimensions of the screen depend on design parameters such as field of view and viewpoint position.
The screen is generally, though not always, rear projected. Three to five projectors are typical. The projectors are mounted above the screen so that the light path is downward to maximize the amount of light delivered to the mirror and, ultimately, to the cabin crew. Since the projection angle is skewed and the projection screen is curved, there is a ferocious amount of geometrical distortion to be corrected by image warping software or special projection lenses. There is also a lot of stray light. Even though the projection volume is light tight, a small percentage of the projected light is reflected from the back surface of the projection screen. Some makes its way to other portions of the screen where it reduces image contrast. Front projected screens are more cumbersome, at least on motion base simulators, but they remove the rear surface reflection problem, are brighter for the same projector power, and offer higher contrast.
A collimated display system is not a 3D display, at least not in the sense that a stereographic system is. However, one effect of collimation is that it reduces perceived binocular depth perception, which in a roundabout fashion, can actual increase the apparent depth of an image and cause the scenery to appear more realistic.
Binocular vision generates strong depth clues from the perspective difference due to eye spacing. These depth clues are, in fact, our most important source of information about the three dimensional nature of our immediate surroundings. Even though a desktop monitor is flat, it is not without binocular depth clues. The slightly different perspective views from each eye are enough for the brainís visual centers to correctly identify the monitor as displaying a flat image 18 inches or so in front of us. The image on the monitor may have all sorts of embedded monocular depth clues for a fantastic flight sim world, but the strength of the conflicting binocular depth clues limits our sense of immersion.
Collimated displays present the same image to each eye with no perspective difference. This doesnít mean there are no binocular depth clues generated, however. If youíve got two functioning eyes, you get binocular depth clues. With the same image presented to each eye, the binocular depth interpretation is that the imagery is distant. Since the imagery is not in the nearby space where binocular depth clue are strongest, the weaker, monocular depth clues become more prominent. While a collimated display may not be a stereographic 3D display, it nonetheless presents an expansive view and allows monocular depth clues to add more to the visual experience.
Thatís a big step to a greater sense of immersion. Iíd like to have one.
Collimated displays are made by Rockwell Collins ( which acquired SEOS in 2008 ), FlightSafety International, and RSI Visual Systems, just to name a few.
Here are links to a few interesting videos: http://www.q4services.com/images/supravue001.mov http://science.discovery.com/videos/...simulator.html
Mike Powell, author of
Building Recreational Flight Simulators and
Building Simulated Aircraft Instrumentation.