3D Television - Technologies - Producing Technologies

Producing Technologies

Stereoscopy is the most widely accepted method for capturing and delivering 3D video. It involves capturing stereo pairs in a two-view setup, with cameras mounted side by side and separated by the same distance as is between a person's pupils. If we imagine projecting an object point in a scene along the line-of-sight for each eye, in turn; to a flat background screen, we may describe the location of this point mathematically using simple algebra. In rectangular coordinates with the screen lying in the Y-Z plane, with the Z axis upward and the Y axis to the right, with the viewer centered along the X axis; we find that the screen coordinates are simply the sum of two terms. One accounting for perspective and the other for binocular shift. Perspective modifies the Z and Y coordinates of the object point, by a factor of D/(D-x), while binocular shift contributes an additional term (to the Y coordinate only) of s*x/(2*(D-x)), where D is the distance from the selected system origin to the viewer (right between the eyes), s is the eye separation (about 7 centimeters), and x is the true x coordinate of the object point. The binocular shift is positive for the left-eye-view and negative for the right-eye-view. For very distant object points, it is obvious that the eyes will be looking along essentially the same line of sight. For very near objects, the eyes may become excessively "cross-eyed". However, for scenes in the greater portion of the field of view, a realistic image is readily achieved by superposition of the left and right images (using the polarization method or synchronized shutter-lens method) provided the viewer is not too near the screen and the left and right images are correctly positioned on the screen. Digital technology has largely eliminated inaccurate superposition that was a common problem during the era of traditional stereoscopic films.

Multi-view capture uses arrays of many cameras to capture a 3D scene through multiple independent video streams. Plenoptic cameras, which capture the light field of a scene, can also be used to capture multiple views with a single main lens. Depending on the camera setup, the resulting views can either be displayed on multi-view displays, or passed along for further image processing.

After capture, stereo or multi-view image data can be processed to extract 2D plus depth information for each view, effectively creating a device-independent representation of the original 3D scene. These data can be used to aid inter-view image compression or to generate stereoscopic pairs for multiple different view angles and screen sizes.

2D plus depth processing can be used to recreate 3D scenes even from a single view and convert legacy film and video material to a 3D look, though a convincing effect is harder to achieve and the resulting image will likely look like a cardboard miniature.