From an imaging point of view, the stereoscopic depth effect can be created using volumetric displays that truly address a projection volume with 3D data information (but lack the solid surface occlusion effect), or using flat LCD screens with extra optical layers in order to address each eye with a separate and credible perspective view. This later approach truly consists in fooling the brain with at least two perspective views that are either computer-generated from 3D data, or recorded as stereoscopic pairs with a multiple-objective camera.
Auto-stereoscopic 3D displays
Commercial 3D LCDs are already available, with notable differences in construction. For example, Sharp has launched a Notebook PC incorporating a 3D LCD monitor. The monitor employs a so-called "parallax barrier" launched at last electronica, where a conventional TFT LCD is combined with a switching LCD. The later establishes an optical parallax barrier between the background lighting and the TFT LCD. This makes it possible to separate left and right halves of stereo pairs that are interleaved on alternate columns of pixels on the LCD, so that slightly different images reach the left and right eyes. When the switching LCD electrically controls the parallax barrier to make it transparent, eliminating its ability to separate light paths, the right and left eyes can see the same image for viewing ordinary 2D content. Horizontal resolution in 3D mode is half that of 2D mode, and because there is only one pair of stereoscopic views, true 3D effect is only perceived within a fairly narrow viewing angle in front of the display and within a specified viewing range.
Dimension Technologies...
Dimension Technologies achieves 3D illusion thanks to a special illumination pattern dubbed "Parallax Illumination", located behind the LCD screen. This makes alternate columns of pixels visible to the left and right eyes. The left image appears on the odd numbered columns and the right image appears on the even numbered columns. The illumination plate, located behind the LCD, optically generates a lattice of very thin, very bright, uniformly spaced vertical light lines that are precisely spaced with respect to the pixel columns of the LCD so that both halves of a stereo pair are displayed simultaneously and directed to corresponding eyes (due to parallax). The company markets a 15.1" and a 18.1" display, the DTI 2015XLS and 2018XLQ respectively, with resolutions of 1024x768 and 1280x1024 pixels.An alternative to parallax barriers is the use of a transparent array of cylindrical lenses overlaying the LCD, pretty much like the old-fashioned ribbed plastic postcards that display 3D images. About the width of three pixels, Each lenticular column modifies the optical path in such a way that the individual pixels behind it are projected to the right and left eyes of the viewer, in coordination with the alternated pixel columns of the right and left stereoscopic views. This approach also offers much better light transmission and improved image contrast for a lower level of background lighting.
SeeReal Technologies...
SeeReal Technologies' 3D displays are based on this principle, using one pair of stereoscopic views. Additionally, to enhance the viewing angle, the company's C-i and C-s 3D LCD displays come with two integrated cameras that provide position tracking (eye tracking for the C-i and spot tracking for the C-s) combined with a mechanism that shifts the lenticular array to adjust the optical path for 3D perception. This frees the user from keeping a locked head position once the 3D image is found, enlarging the 3D viewing angle to ±20°. The displays offer resolutions up to 1280x1024, that is 640x1024 per eye, for a viewing distance of 55 to 75cm. The C-nt, a 1600x1200 pixels 20" 3D TFT display is under development.
Multiple viewers
In order to offer multiple-viewers capability and enhance the 3D perception with motion parallax (look-around effect when moving the head position respective to the screen), multiple pairs of stereoscopic views must be used, while maintaining at any time, only one stereoscopic pair visible for any one user in front of the screen. Philips Research has worked in that direction with a lenticular array implementation. The unique feature of the Philips 3D-LCD prototype is that the lenses are not vertically aligned with the columns of the LCD, but are placed at a slight angle. With this optical arrangement, the LCD pixels are sampled as a function of the view angle, and up to seven views (forming 6 stereoscopic pairs) are interleaved into a complex pixel pattern that matches the optical sampling. The positioning of adjacent lenses is such that the pixel images are interspersed rather that purely overlapping. This creates a continuous viewing zone without gaps, a smooth transition of fading views that achieves the illusion of a solid object rather than a succession of discrete views. The odd numbered views are imaged by the first lens and the even numbered views by the second lens. With sideways movement, a user comes across different stereoscopic pairs (for example the left and right eyes see view number 5 and 3, and then moving to the right, the left and right eyes views change simultaneously to 4 and 2). While allowing unrestricted head movement in front of the display without loosing the 3D illusion, the seven views provide continuous motion parallax and because the viewing zones repeat with sideways movement (i.e. ...3,4,5,6,7,1,2,3,4,5,6,7,1,2,3...) the display can be viewed by multiple users at the same time. A 14.5" prototype has been built to prove the concept, in cooperation with Hosiden Display Corporation, based on a 1024x768 colour display, offering a resolution per view of 438x256. One of the key advantages of the slanted lenticular approach is that both ththe horizontal and vertical pixel count in the LCD is used to trade-off against the number of views. Viewing distance ranges from 30 to 150cm.
StereoGraphics...
Reaching market with the SynthaGram monitors, StereoGraphics also uses proprietary lenticular sheets located in front of its LCDs. The lenticular optics are tipped from the vertical and mapping of the pixels is accomplished at the sub-pixel (red, green, and blue) level to interleave up to nine different views (eight stereoscopic pairs). Image pixels of the appropriate perspectives are directed to each eye over a wide viewing angle and repeated in five symmetrical viewing zones (of 20° span each) for multiple users and viewing comfort. To preserve the precise mapping of the pixels and their association with the optical elements, a video accelerator board can be provided with a digital video interface. What's more, any number of perspective views, two or greater, may be used (with different pixel mappings). The 22.2" SynthaGram 222 has 3840x2400 of 2D resolution, and about one third the resolution per eye in 3D mode.
Deep Light...
Rather than relying on LCDs, Deep Light's rear projection 3D display consists of a display engine integrated at the back of the display housing, and projecting multiple stereoscopic images directly to the viewers' eyes (via a mirror-based optical path). The illusion of 3D is rendered directly without any optical elements interfering. What's more, by driving the display engine in two slightly different angular directions, the display is capable of providing two different video contents at full 1280x1024 resolution to two different users sitting side by side (they only need separate audio headsets to completely ignore each other, or a directional sound option is also available). This side by side viewing option is available in both 2D and 3D modes.
Holografika...
The Hungarian firm Holografika E.C markets the HoloVizio 128w, a 32" display featuring 10 million pixels, 24-bit colour resolution and a 50Hz refresh rate. The HoloVizio technology is based on hologram optical geometry principles. Thanks to a specially arranged array counting many spatial light modulators and optical mirrors, each volumetric pixel of the holographic screen is able to emit up to 60 light beams of different colour and intensity in a 50° angular viewing zone, corresponding to a real object perception as the viewers move in front of the screen. Proper software control makes the light beams leaving the various pixels to propagate in multiple directions, as if they were emitted from a common object point behind the screen, or the beams can cross each other in front of the screen. In this way, viewers will perceive the points as objects behind the screen or floating in the air in front of the screen, respectively, providing the depth and occlusion effect of holograms, with horizontal-only parallax. The system reconstructs the key elements of spatial vision, similarly to holographic stereograms, hence eliminating the need of handling enormous amounts of redundant information associated with purely holographic systems (that typically compute light interference fringe and diffraction patterns from 3D ray tracing).
And here comes 3D TV
For pure virtual 3D content in medical or engineering applications, generating multiple views from 3D data is only a matter of efficient computing. But for a broader market acceptance of 3D displays, and ultimately to reach the TV market (with high-definition 3D-enhanced garbage content), traditional 2D content creators must be given a chance to exploit the auto-stereoscopic capabilities of these displays, even before costly stereoscopic live-recording becomes the dominant process. In this optic, a number of companies including Philips Research, StereoGraphics, or Dynamic Digital Depth offer to convert existing 2D video footage into 3D viewable content. This is done by running sophisticated software that extracts depth information from 2D shadows and motion characteristics and then provide the equivalent stereoscopic views in the desired display format. Dynamic Digital Depth's proprietary tool for 2D to 3D conversion generates depth maps and encode them within the 2D video stream for broadcasting. The encoded image is decoded at the viewpoint and watched in a variety of 3D formats chosen by the viewer or in its original 2D state when using conventional televisions or PCs. This conversion process is very compute-intensive and currently involves some semi-manual processes. The company also offers to adapt third-party professional visualization programs to any 3D format on the market, and develops 3D display drivers so that any 3D content input can be mapped to the correct pixel patterns unique to each 3D display. What's more, with these software solutions, 3D depth intensity can be adjusted to user's comfort level.
3D Consortium...
To further encourage the widespread application and expansion of 3D technology, five Japanese companies, namely Itochu, NTT Data, Sanyo Electric, Sharp, and Sony have created the 3D Consortium, in March of last year. Already strong of over 70 members, the 3D Consortium encompasses hardware manufacturers, software vendors, content providers, broadcasters or even academic organisations. The consortium plans educational campaigns and will be dealing with a number of specific issues such as spreading image formats appropriate to various applications and I/O devices, and developing guidelines and authoring tools for creating content. Other companies such as Samsung, LG, NEC-Mitsubishi Electronic Displays or Toshiba are also active in this field, with new products showcased in October 2003 at last CEATEC (Combined Exhibition of Advanced Technologies) in Japan.
