With this technology, certain groups of pixels on the screen are only visible within a particular range of angles with respect to it, each thereby projecting a different perspective view. In order to achieve this effect, two following parts of technology must function together. First of all, the software feeding data to the display panel must multiplex the images corresponding to all the perspective views into one composite image. This image viewed directly does not have any depth and looks like a blurry version of a conventional flat picture. Secondly, an optical system designed with the particular dimensions and pixel pitch of the display panel, as well as with the number and angles of projected views in mind must be placed in front of the panel. This system ensures that the light form pixels (or, in most implementations, rather colour sub-pixels) belonging to the same perspective view is only visible from the corresponding viewing angle (see Fig. 1).
Fig. 1. 9-view space-shared 3D display using an HD flat panel.
In one of its simplest forms, such a system may contain a single sheet of lenticular lenses whose pitch depends on the display pixel pitch and on the number of perspective views, and whose lenses are positioned vertically to provide horizontal parallax. Unfortunately, this configuration has a critical flaw: the horizontal resolution of such 3D display would be divided by the number of views while its vertical resolution would remain the same as that of the underlying light panel, leading to visual artefacts whose severity is proportional to the number of views.
In order to balance out the loss of horizontal resolution, a slightly slanted (usually at ≈ 9½° to the vertical direction) lenticular sheet is used (see Fig. 2 for more details). With correctly chosen relationship between the flat panel resolution and the lenticular sheet parameters, this leads to evening out of the resolution between the horizontal and the vertical dimensions.
Fig. 2. Detail of 9-view space-shared 3D display using a slanted lenticular sheet.
Current offerings using flat panel screens (for example, 3D Fusion, Fujitsu, Toshiba, etc.) appear to achieve this by using lenticular lens arrays, parallax barriers (periodic masks), and/or some form of micro lens arrays at the front of a 2D panel. However, even with the horizontal and vertical resolution evened out, this approach relies on the distribution of the total number of pixels in the screen among all the views, so that the pixel density of the resulting 3D image is effectively divided by the number of views projected. Since resolution is a function of pixel density, this means that high-quality 3D images with the resolution equivalent to that of conventional 2D displays would only be possible with flat panels whose pixel density is multiplied by the number of views a 3D display is designed for. Glasses-free, flat panel displays presently allow for up to 9 views. These views repeat several times over the entire viewing angle of the screen, forming separate viewing zones from which the identical image can be observed. This means that the base HD resolution has to be divided by 9 for each view. In addition, the head movement in any one zone is restricted to only few centimetres and the viewing distance from the screen may also be restricted in order to achieve the 3D effect.
Even with only 5 views, in order to achieve the full HD playback (1920 × 1080) for each view, the total screen resolution must be 5 times HD, or approximately 9.5 million pixels (4096 × 2304). To provide for wide-angle and contiguous 3D view ability along with an adequate viewing angle for multiple viewers (> 40 views), it is estimated that approximately 83 million pixels are required, or pixel densities more than 40 times those currently available. This sort of technology does not presently exist and even if it were otherwise then manufacturing costs would likely be prohibitive. At present, some industry players are working on the development of the so-called UHDTV (Ultra High-Definition TV) with the resolution of up to 7680 × 4320 pixels. This is expected to become available in 2022. Even with such extremely high native resolution, the maximum number of views in HD 3D would be a meagre 16.
It is appropriate here to make some clarifications on terminology. One should not confuse number of “views” (perspectives) with number of “viewers” (individual simultaneous observers).
The number of views is the number of perspectives from which a 3D object can be observed within specific and different angles. Multiple-view auto-stereoscopic 3D displays create many perspectives, giving not only the sensation of volume but also freedom of movement for one or many observers. The more perspectives (views), the larger the viewing zone and the more freedom of movement the observer has. The size of the viewing zone is directly proportional to the width of each view at the observer’s distance from the display. But this width cannot exceed the inter-pupillary distance (the average distance between the pupils of the observer’s eyes), otherwise the observer loses the 3D perception. Views are also correlated to the viewing distance. Number of viewers means the number of observers who can watch 3D images on a 3D display simultaneously.