Sunday, July 01, 2012 1:47 AM
Creating the next generation of computer graphics 3-D
When someone looks at a computer screen straight on they can see the perception of a 3D environment. But when the same person looks at the image from the left or right of center the image still appears flat. To make the image have depth and 3D perhaps stacking pixels on top of each other in squares would be the best way to make the image 3D.
The image would be mapped on each pixel so that each pixel segment of the image wraps around the pixel on all five sides of the pixel.
The front side of the pixel would represent the image when looked straight on.
The left and right side of the pixel would represent the image when viewed from the left or right. The top and bottom side of the pixel would represent the image when looked at from the top or bottom.
If the viewer was looking at the top of a womans head by standing up and looking down on the screen the pixels would adjust theirself to show the top of her head and what was behind her thus creating a real 3D image.
Sunday, July 01, 2012 7:54 PM
Cute idea, but this will not lead to real 3D.
Imagine a window with pixels and let us examine just one pixel on that window. For a flat, 2D image, this pixel has just one value. To represent real 3D, the pixel must show the value that would be seen from each view angle in front of the screen. When viewed from some position by a human observer with two eyes, the pixel must show the two values that the eyes would see when looking through a window. As the viewer moves there will be different values for each position. To produce real 3D then, each pixel must be able to present an infinate number of values - or a reasonable approximation.
Holograms come close to doing this. For 3D movies and 3D TV just two sets of values are presented: the images as seen by two "eyes" from just one observing position. When the viewer moves these images do not change (assuming that the viewer is wearing 3D glasses to sort the left image to the left eye, and the right image to the right eye).
The two-image approach is technically feasible and relatively easy, but requires glasses for sorting out the images.
It is also possible today to substantially increase the number of actual images presented at the cost of lower resolution. There is a camera available that tries to do just that for eliminating the need to focus, but not for 3D.
It would take a lot more than five sets of images for any reasonable real 3D. However, with just 5 images fairly good real 3D can be resonstructed for two eyes withing a small horizontal view angle (say 30 degrees) by computing "tween" images and presenting them with a lenticular screen. The closest to this that has been done was the Nimslo camera and prints, the camera resorded four images.
Sunday, July 01, 2012 8:56 PMOwner
Here is an example of what a 3D display might look like: http://www.youtube.com/watch?v=YKCUGQ-uo8c&feature=player_detailpage#t=26s
As Ludwig points out, this requires 288 different viewpoints to be generated simultaneously in order to allow a full 360 walk-around.