Frensnel lens collimation question?

[mikesblack]

Perhaps this has been discussed more in another thread, but I haven't found it yet. I was wondering about head tracking and collimated optical view. Has there been anyone who has found a way to make this work? I understand Track IR will only work with one window and not decoupled multiples as per Wide View or fast PC 3 windows etc. on a curved screen. I believe one FSX window provides only 120 degreed?? Perhaps, a forward view with head tracking for depth view and runway alignment and fixed views from the other image generators? Not perfect, but I wonder if it would be a good enhancement. I plan on building a spherical screen of 220 degrees and N Thusim. Any ideas on head tracking or other relavent info would be greatly appreciated.

[Neil Hewitt]

Here's a thought - if you used the nVidia shutter-glass 3D system to view FSX in 3D (which is possible, the driver will support it), would the outside view be effectively collimated, with the 3D perspective tricking your eye into seeing things at the 'right' focal length?

Apparently there are some problems with texture and lighting effects when using FSX in 3D, but nothing that's insoluble. A (relatively) cheap pair of glasses would let you get the equivalent of an expensive collimated display while still seeing the cockpit interior etc unchanged.

Anyone try this? I'm tempted to go out and buy the glasses now just to experiment...

[Mike.Powell]

In a collimated display the image is moved far from the viewpoint with several results:

The left eye and the right eye see the same image with no perspective difference. It may seem that viewing a non-stereographic monitor would present the same image to each eye, but that's not so. Because the eyes are about 2.5 inches apart, each sees a different, slightly skewed view of the monitor surface. This small difference has a huge impact on what's perceived. Although the pixels are the same, the perspective difference is enough for the visual center in the brain to say, "Those pixels are painted on a flat screen about 18 inches away". When a collimated display is viewed, that same visual center says, "Those pixels aren't very close. They may be flat, or maybe not".


Because a collimated display moves the image away, the eyes focus for distance vision, and they point straight ahead.

When the viewer moves sideways, the image appears to move with him. This is the difference between looking at your monitor on your desk and moving your head a few inches sideways, and looking out the window and moving your head. In the first case the monitor and image on the monitor begin to move out of your field of view. In the second, the window frame begins to move out of your field of view but the outside view seems to be stationary. A collimated display presents an image that acts more like a window and less like a desktop monitor.

So, if you wanted to get the same effects as a collimated display, perhaps you could...

Develop a driver that generates imagery for each eye that removes the normal left/right perspective difference of viewing the monitor on your desk. This is actually an "anti-3D" driver. It's not putting binocular 3D visual clues into the image. It's making sure there aren't any. Use the LCD shutter glasses for viewing.

Use a head tracker to change the viewpoint dynamically. When you move your head relative to the monitor the image shifts as though the monitor were a window frame.

There are few limitations. The eyes will still focus on a screen that is close by. Initially this may feel uncomfortable because they will not be rotated inward slightly as they normally would be when viewing close images.

Also, this approach would work for only a single viewer at a time.


While I've read of an application that varies the displayed image based on viewer position to simulate a window to the outside, I don't know of anyone doing the "anti-3D". I think it would be a very interesting experiment.

[castle]

It might produce the desired visual effect if the OTW scene was a large screen projection 3 or 4 feet in front of the viewer bounded by the frame of a cockpit window. But it would not solve the perspective problem of side by side seating.

The optics of collimated and 3D displays are different. But it is certainly cheaper and worth a try.

Have not messed with 3D systems. Would it work for a multi-screen display and is there sufficient horsepower ( video pipeline, GPU, CPU ) to provide a minimum frame rate suitable for stutter free viewing.

JW