3-D Displays: A review of current technologies
mUltimo3D Home
3-D Visual OS
3-D Display
Head Tracker
Gaze Tracker
Hand Tracker
mUltimo Staff
mUltimo3D is a registered project of the World Exposition Germany


Siegmund Pastoor and Matthias Wöpking

Ongoing developments in information technology have led to an increasing demand for 3-D displays. Indeed, a broad range of fairly mature 3-D equipment is already on the market. The available systems, however, suffer from the drawback that users have to wear special devices to separate the left eye's and right eye's images. Such "aided viewing" systems have been firmly established in many professional applications. Yet further expansion to other fields will require "free viewing" systems with improved viewing comfort and closer adaptation to the mechanisms of binocular vision. The respective technologies are still under development. This article reviews the state-of-the-art of both aided viewing and free viewing systems.

Keywords: 3-D displays, stereoscopy, spatial vision

The proliferation of 3-D technology is currently gaining impetus from a number of activities. The driving impact originates not only from the traditional entertainment sector (television and cinema), but is also fed by many new sources: content providers in the multimedia field have spotted a market for true 3-D games and adventure rides in virtual reality, software-developers have recognized the opportunity to launch a new generation of three-dimensional graphic user interfaces, the Internet community is specifying data structures for the transmission of 3-D objects (the Virtual Reality Modeling Language), and EC-sponsored R&D programs (RACE and ACTS) are aimed at preparing the infrastructure for 3-D videoconferencing. Moreover, the base technologies for a future 3-D market have made considerable progress. As will be shown by various examples, 3-D display development has profited enormously from recent advances in LC-technology. Other factors which will play a leading role are the new programmable multimedia signal processors, the rapid growth of highspeed digital networks and the supply of ever more computing power at low costs.

Therefore, the chances for success are high, if and when 3-D displays become available which enable immediate space perception without any impairing side effects. This is the crucial technological challenge. Appropriate solutions require a thorough understanding of the mechanisms of binocular vision. The next section will outline the relevant perceptual processes. This is followed by a taxonomy of 3-D display techniques. The main part of the paper reviews the presently known forms of implementation.


Spatial vision is based on monocular and binocular depth cues contained in the scene observed. Monocular cues are visible with one eye and are used to create an illusion of volume and depth on flat image surfaces. Common examples are linear and aerial perspective, texture gradients, overlap of objects, and the distribution of light and shadows.

Binocular depth perception is based on local displacements between the projections of a scene onto the left and right retina (disparity). As long as these displacements do not exceed a certain magnitude, the two projections are merged into a single percept with structured depth (fusion). This condition is safeguarded by the operation of several interacting mechanisms:

Depth-of-focus is an integral constituent of spatial vision. Automatic focusing of the eye's lens (accommodation) makes the currently fixated object stand out against its surroundings; details in the near foreground and in the far background "disappear" in increasing optical blur. This way, disparity is kept within physiological limits. In order to maintain a moving object within the zone of sharpest vision, the eyeballs perform rotational movements (convergence). Thus, accommodation and convergence are used to carrying out coordinated changes. As a consequence, all objects which are in focus are also within the limits of fusion - independent of their absolute distance from the observer.

Most of the currently feasible 3-D display techniques neither support the gliding depth-of-focus nor the linkage between accommodation and convergence (Fig. 1). This failure is a potential cause of visual strain, and generally limits the comfortable viewable depth volume [1].



      Fig. 1: 3-D displays usually force the observer to  "freeze" accommodation in the screen plane, irrespective of the fixated object's  distance. Consequently, all details of the scene appear within unvarying depth-of-focus.  Normally, accommodation distance corresponds to convergence distance and the  depth-of-focus range is centered at the fixated object.


Taxonomy of 3-D displays

3-D displays can be categorized by the technique used to channel the left and right images to the appropriate eyes (Table 1): some require optical devices close to the observer's eyes, while others have the eye-addressing techniques completely integrated into the display itself. Displays of the latter category are called autostereoscopic and are technically much more demanding than the type with viewing aids (stereoscopic displays).

Table 1 Classification of 3-D display techniques.

Principle of eye addressing

Effective  origin of waves

Number  of different views

Eye-point  dependent perspective

aided viewing (stereoscopic)

color multiplex,
polarization multiplex, time multiplex,
location multiplex

image plane, gaze-controlled
image plane


(for a single observer)

free viewing

direction multiplex (e.g., by diffraction, refraction, reflection, occlusion)

image plane, gaze-controlled
image plane

³ two

(for a small number of observers)


volumetric display, electro-holography

distinct depth planes  (slices), entire space


(for a small number of observers)


With stereoscopic displays two different perspective views are generated (quasi-)simultaneously. The waves of light entering the eyes may have their origin on a common, fixed or movable (gaze-controlled) image plane. Several multiplexing methods have been proposed to carry the optical signals to the appropriate eyes. It is possible to adapt the image content to the current head position (eye-point dependent perspective; e.g. via corresponding camera movements or signal processing).

With autostereoscopic systems, the only exploitable constraint for addressing the left and right eye is the fact that they occupy different points in space. The waves of light forming the 3-D image may originate from fixed or gaze-controlled image planes. In both cases, direction-multiplex is the only way to channel the information of the left and right views into the appropriate eyes. Compared to stereoscopic techniques, it is possible to multiplex more than two views at a time. Thus, individual (eye-point depending) perspective views can be delivered to different observers. The volumetric and the electro-holographic approaches produce 3-D images where the effective origin of the waves entering the observer's eyes match with the apparent spatial position of the corresponding image points. Thus, the fundamental mechanisms of spatial vision are perfectly supported (in principal there is no difference to natural viewing conditions).

Stereoscopic techniques

Current approaches to 3-D displays are described in detail in the following sections, starting with stereoscopic techniques. Except for a recently proposed location-multiplex approach, all stereoscopic techniques force the observer to focus on a fixed image plane (Fig. 1).

Color-multiplexed (anaglyph) displays

In anaglyph displays the left and right-eye images are filtered with near-complementary colors (red and green, red and cyan or green and magenta, and the observer wears respective color-filter glasses for separation. Combining the red component of one eye's view with the green and blue components of the other view allows some limited color rendition (binocular color mixture). Color rivalry and unpleasant aftereffects (transitory shifts in chromatic adaptation) restrict the use of the anaglyph method.

Polarization-multiplexed displays

The basic setup consists of two monitors arranged at a right angle and the screens are covered by orthogonally oriented filtersheets (linear or circular polarization). The two views are combined by a beam combiner (half-silvered mirror) and the observer wears polarized glasses. This frequently used technique provides full color rendition at full resolution and very little cross-talk in the stereo pair (less than 0.1% with linear filters). However, over 60% of the light emmited is lost through the filters, and the remaining light flux is halved by the mirror.

Polarization techniques are very well suited for videoprojection. When using CRT-projectors with separate optical systems for the primary colors, the left and right-view color beams should be arranged in identical order to avoid rivalry. The light flux in LC-projectors is polarized by the light-valves. Commercial LC-projectors can be fitted for stereo display by twisting the original polarization direction via half-wave retardation sheets to achieve, e.g., the prevalent V-formation.

Stereo projection screens must preserve polarization. Optimum results have been reported for aluminized surfaces and for translucent opalised acrylic screens. Typical TV rear-projection screens (sandwiched Fresnel lens and lenticular raster sheet) depolarize the passing light. LCD-based, direct-view displays and overhead panels have recently been marketed [2]. Their front sheet consists of pixel-sized micro-polarizers, tuned in precise register with the raster of the LCD. The left and right-eye views are electronically interlaced line-by-line and separated through a line-by-line change of polarization.

Time-multiplexed displays

The human visual system is capable of merging the constituents of a stereo pair across a time-lag of up to 50 ms. This "memory effect" is exploited by time-multiplexed displays. The left and right-eye views are shown in rapid alternation and synchronized with an LC-shutter, which opens in turns for one eye, while occluding the other eye. The shutter-system is usually integrated in a pair of spectacles and controlled via an infrared link. When the observer turns away from the screen, both shutters are switched to be transparent. Time-multiplexed displays are fully compatible for 2-D presentations. Both constituent images are reproduced at full spatial resolution by a single monitor or projector, thus avoiding geometrical and color differences. The monitor-type systems have matured into a standard technique for 3-D workstations.

Impairing cross-talk may result from the persistence of CRT phosphors, particularly of the green phosphor. Ideally, image extinction should be completed within the blanking interval. Using "fast" phosphors like P46 leads to noticeable loss of brightness (50%) and life-span. P43 yields a somewhat optimum performance. With the standard phosphor P22, cross-talk can amount to over 20% (the perception threshold is at 0.3% [3]). Since the two views are displayed with a time-delay, they should be generated with exactly this delay. Otherwise, moving objects will appear at incorrect positions in depth.

Time-sequentially controlled polarization

Tektronix has developed a display which combines the time and polarization-multiplex techniques. The monitor's faceplate is covered with a modulator, consisting of a linear polarizer, a liquid-crystal p -cell and a quarter-wave retardation sheet (to turn linear polarization into circular polarization). The p -cell switches polarization in synchronism with the change of the left and right-eye views. Circular polarizing glasses serve for de-multiplexing.

This approach offers three advantages over systems with active shutter-glasses: First, polarizing glasses are inexpensive and light-weight. Second, the p -cell can be constructed from several segments which operate independently on the active portions of the screen, hence ensuring that each eye is only reached by the intended image contents. This way, cross-talk can be greatly reduced (down to 3.6% in a system with five segments and the P22 standard phosphor [4]). Third, multiple display arrays can be operated without any extra synchronizing circuitry.

Location-multiplexed displays

Location multiplex means that the two views are created at separate places and relayed to the appropriate eye through separate channels (e.g. by means of lenses, mirrors and optical fibres). In the following examples, the image plane appears either in a fixed accommodation distance (HMD and BOOM displays) or in a variable, gaze-controlled position (3DDAC).

Helmet Mounted Display (HMD): With HMDs, the perceived images subtend a large viewing angle, typically up to 120° (horizontal) by 80° (vertical). As the natural surroundings are occluded from sight (sometimes optional, by a visor), HMDs are apt to convey a feeling of total immersion in the scene displayed. Attached headtracking devices are used to create eye-point dependent changes in perspective when the user moves. Accommodation distance is usually below 2 m, or can be adjusted in see-through displays to the distance of the fixated real world objects [5]. Meanwhile, numerous manufacturers offer a variety of systems with different performance [6].

HMDs allow free head movement without losing screen contact, thus avoiding musculo-skeletal problems. There is no need to dim ambient light (visor closed), and peripheral vision must not necessarily be obstructed (visor open). On the other hand, latencies and tracking-errors tend to provoke odd, uneasy sensations through conflicting visual stimulation and postural feedback, and adaptation can lead to reciprocal aftereffects. A disadvantage of see-through HMDs is that the observer must direct his regard towards a dim background in order to perceive the HMD image.

BOOM-Display: The Binocular Omni-Orientation Monitor (BOOM [7]) was designed to release the user from the encumbrance of wearing an HMD. Two miniature CRTs are accommodated in a case, which is mechanically supported by a counterweighted, six degrees of freedom arm. The observer regards the stereo images as if looking through binoculars. Tracking is implemented by optical shaft encoders at the joints of the boom. The monitor case is moved either with hand-held handles or with the head (like an HMD).

3DDAC-Display: ATR Labs (Kyoto) have developed an HMD concept with movable relay lenses interposed between the screen and the eyepiece lenses [8]. The convergence distance is sensed by a gaze-tracker. The position of the relay lenses is constantly adjusted, so that the screen surface (image plane) appears in the same distance as the convergence point of the gaze-lines (Fig.2). Thus, the "3-D display with accommodative compensation" (3DDAC) supports the natural link between accommodation and convergence.




      Fig. 2: Location-multiplexed display with gaze-controlled image plane


Autostereoscopic techniques

The present situation regarding autostereoscopic displays is marked by a variety of competitive approaches. Therefore, this survey mainly focuses on fundamental trends. The range of currently recognized techniques is presented in the following order: electro-holography, volumetric displays and direction-multiplexed displays.


Holographic techniques can record and reproduce the properties of light waves - amplitude (luminance), wavelength (chroma) and phase differences - almost to perfection, making them a very close approximation of an ideal free viewing 3-D technique. Recording requires coherent light to illuminate both the scene and the camera target (used without front lenses). For replay, the recorded interference pattern is again illuminated with coherent light. Diffraction (amplitude hologram) or phase modulation (phase hologram) will create an exact reproduction of the original wavefront.

Video-based holographic techniques are still in their infancy, although they have received much attention over the past five years. Organised by TAO (Telecommunications Advancement Organization), several Japanese research institutes are working towards adapting the principle of holography to an LCD-based video electronics environment [9]. However, the spatial resolution of today's LC-panels is a serious bottleneck (minimum requirements are 1000 lp/mm). A possible solution is to partition the hologram among several LC-panels and to reassemble the image with optical beam combiners. It is yet an open question how to store and transmit the enormous amount of data contained in a hologram. The source rate is estimated to exceed 1012 bit/sec [10]. Specific data compression methods are required [11]. Currently, the scope of this approach is limited to very small and coarse monochromatic holograms (width of field 1 cm [12]).

Holograms cannot be recorded with natural (incoherent) lighting - a decisive shortcoming. Therefore, they will remain confined to applications where the scene is available in the form of computer generated models. An approach based on computer generated holograms has been pursued since the late 1980s at MIT's Media Lab [13].

The MIT approach makes intense use of data reduction techniques (elimination of vertical parallaxes, subsampling of horizontal parallaxes). Yet, the pixel rate for a monochrome display with a diameter of 15 cm, a depth of 20 cm, and a viewing zone of 30 degrees amounts to 2 Gigapixel/sec (at 30 Hz frame rate). The hologram is displayed with acousto-optical modulators (tellurium dioxide crystals transversed by laser light, Fig. 3). At any one instant, the information of nearly 5000 pixels travels through a crystal. For optical stabilization, synchronized oscillating mirrors are interposed between the modulator and the observer's eyes. Vertical sweep is accomplished with a nodding mirror. Current work aims at increasing the image diameter to 25 cm by assembling multiple basic elements in parallel.




      Fig. 3: MIT holographic display. Light modulation is effected by  frequency-modulated soundwaves which propagate through the acousto-optical modulator and  cause local changes of the refractive index due to pressure variations. Phase shifts in  the lightwaves cause local retardations used to create a phase hologram.


Volumetric displays

Volumetric displays project image points to definite loci in a physical volume of space where they appear either on a real surface, or in translucent (aerial) images forming a stack of distinct depth planes. With the first type of system, a self-luminous or light reflecting medium is used which either occupies the volume permanently or sweeps it out periodically. Technical solutions range from the utilization of fluorescent gas (with external excitation through intersecting rays of infrared light) over rotating or linearly moved LED-panels to specially shaped rotating projection screens. Rotating screens have been implemented in the form of a disc, an Archimedian spiral or a helix, winding around the vertical axis [14].

Among the displays with a real collecting surface, the helical design has reached maturity. The most elaborate equipment uses a double-helix filling an 91-cm-diameter by 46-cm-high volume at 10 revolutions per second with a maximum of 40,000 color pixels per frame (or 120,000 pixels in the color primaries) [15]. Observers can walk around the display and see the imaged objects from different angles.

The second type creates aerial images in free space which the observer perceives as cross sections of a scene lined-up one behind the other (multiplanar display). The images belonging to different depth layers are written time-sequentially to a stationary CRT. The observer looks at the screen via a spherical mirror with a varying focal length (varifocal mirror [16]). Rendition of the depth layers is synchronized with changes of the mirror's shape, so that slices of the 3-D volume are successively created in planes of varying distance. The oscillating process is repeated at 30 Hz (which is insufficient to avoid flickering). Due to phosphor persistence, only a limited number of planes can be displayed without visible image smear.

Usually, the mirror is a circular flexing membrane with a metalized surface which is forced to oscillate by means of an acoustical woofer. The deformation of the mirror (maximum 4 mm) yields an optical leverage of approximately 70:1. Variations of optical magnification are compensated by reciprocal magnification of the CRT-images. A volumetric display with additional display of high-definition 2-D background images has been recently announced [17]. In volumetric displays the portrayed objects appear transparent, since the light energy addressed to points in space cannot be absorbed by foreground pixels. Practical applications seem to be limited to fields where the objects of interest are easily iconized or represented by wireframe models.

Direction-multiplexed displays

Direction-multiplexed displays apply optical effects like diffraction, refraction, reflection and occlusion in order to direct the light emitted by pixels of different perspective views exclusively to the appropriate eye. Recent technical approaches outlined in the rest of the paper make use of this concept.

Diffraction-based approaches

DOE approach: With the diffractive-optical-elements (DOE) approach, corresponding pixels of adjacent perspective views are grouped in arrays of "partial pixels" (ICVision Display [18] and 3D Grating Image Display [19]). Small diffraction gratings placed in front of each partial pixel direct the incident light to the respective image's viewing area (first order diffraction; A1 to A4 in Fig. 4). Current prototypes yield images of less than 1.5 inches in diameter. Advanced concepts provide for the integration of image modulation and diffraction of light within a single, high-resolution spatial light modulator [20].




      Fig. 4: Principle of a DOE 3-D display (after [19]).


HOE approach: Holographic optical elements (HOE) model the properties of conventional optical elements (e.g. lenses) by holographic methods. Thus, an HOE contains no image information, but serves to diffract rays of light modulated elsewhere. In a recent prototype display, the HOE is an integral part of the light box of a modified LCD [21]. It consists of the hologram of an even diffuse plane which is rastered so as to direct the light of alternating lines to specified viewing zones. Outside the zone for stereoscopic viewing, both eyes receive a 2-D view. The stereo zone can be made to follow the observer's head movement by moving the light source.

Making the above HOE takes two exposures of a hologram with the same illumination setup. Between the exposures, the object, which in this case is a diffuse plane, is shifted horizontally by its own width. For each exposure, parts of the hologram corresponding to the odd or even image lines are occluded.

Refraction-based approaches

Numerous display concepts have been proposed based on conventional, refractive optical elements (such as picture-sized large lenses or small lenslets) to address the observer's eyes. These concepts are discussed in the following.

Integral imaging: With integral imaging, the spatial image is composed of multiple tiny 2-D images of the same scene, captured with a very large number of small convex lenslets (fly's-eye lens sheet). Each lenslet captures the scene from a slightly different perspective. A lens sheet of the same kind is used for display (between capture and replay, the image has to be inverted for orthoscopic depth rendition [22]). As the image plane is positioned into the focal plane of the lenslets, the light from each image point is emitted into the viewing zone as a beam of parallel rays at a specific direction. Therefore, the observer perceives different compositions of image points at different points of view.

The individual lenslets must be very small, since each pixel is spread to the lens diameter at replay, and the image formed behind each lenslet should be as complete and detailed as possible. As a consequence, the display must provide an extremely high spatial resolution. Basic research on electronic video equipment for integral imaging is being carried out at De Montfort University (Leicester) [23].

Lenticular imaging: Lenticular techniques use arrays of vertically oriented cylindrical lenslets, and can be considered a one-dimensional version of integral techniques. The light from each image point is emitted at a specific direction in the horizontal plane, but non-selectively in the vertical plane. Therefore, changes of perspective in accordance with vertical head movements cannot be achieved by optical means (but by headtracking and computational image processing). Based on lenticular techniques, direct-view and projection-type, 3-D displays have been implemented [24].

The working principle of a direct-view display is shown in Fig. 5. Two stereo half images are presented simultaneously, with two columns of pixels (one for the left and one for the right-eye image) behind a lenslet. The observer's head position is constantly sensed, and the lenticular sheet is shifted mechanically to track any lateral and frontal movements. Because of the cylindrical lenses' horizontal selectivity, the imaging panel's color-filter stripes must be aligned one above the other. Otherwise, the observer would see separated color components from different view points.


      Fig. 5: Direct-view lenticular raster display with headtracking. L  and R denote corresponding columns of the left and right-eye images.

Headtracking can also be achieved by moving the complete display to and fro and rotating it around its vertical axis [25]. Alternatively, a large number of pixel-columns of the different perspective views has to be displayed simultaneously behind each lenslet [26]. The principle of a purely electronical headtracking display is shown in Fig. 6 [27]. An LC-shutter with movable vertical slits is placed in the focal plane of the lens sheet, and the left and right-eye views are time-sequentially displayed on a CRT screen. The slit position is switched in sychronism with the change of views and shifted laterally according to the observers head position.


      Fig. 6: Direct-view display with purely electronic headtracking.

Fig. 7 shows the principle of a rear projection 3-D display using a stereo projector and a dual lenticular screen [28]. The rear screen serves to focus the projected images in the form of small vertical stripes onto a translucent diffusor. The lenses of the front screen map the vertical image stripes to specific loci in the viewing zone - quasi mirroring the initial path of light during rear projection. Since the image stripes on the diffusor also pass through a set of adjacent lenslets, there are hence several adjacent viewing zones. Headtracking is possible by corresponding shifts of the front lens sheet (same principle as in Fig. 5), or by moving the stereo projection unit on a mirror-inverted path. In the latter case, multiple user access systems would need to operate several independent pairs of projectors [29]. Lenticular 3-D displays with headtracking were not constructed until very recently. In the past, multiview capability was approached by a large number of stationary projectors (up to 40 [30]).


      Fig. 7: Principle of a rear projection display with dual lenticular screen

Manufacturing dual lenticular screens is a rather delicate process, since the two raster sheets must be brought into precise register and the whole surface must be of perfect homogeneity. These problems are somewhat attenuated with front projection systems (single raster sheet with reflective rear coating), since identical optical elements are used to write and read the image [24].

It is also possible to use a single projector in order to rear-project rastered image stripes directly onto the diffusing plate behind the frontal lens sheet. In this case, headtracking can be achieved by small horizontal shifts of the rastered image (e.g., by shifting the projector's front lens). Frontal movements are followed up by adjusting the magnification of the rastered projection image. Again, an alternative for tracking is simultaneous projection of multiple rastered views, with a respective the number of image stripes accommodated behind each lenslet [31].

Field-lens displays: A field-lens is placed at the locus of a real (aerial) image in order to collimate the rays of light passing through that image, without affecting its geometrical properties. Various 3-D display concepts use a field lens to project the exit pupils of the left and right-image illumination systems into the appropriate eyes of the observer. The effect is that the right-view image appears dark to the left eye, and vice versa. This approach generally avoids all the difficulties resulting from small registration tolerances of pixel-sized optical elements.

The basic principle is shown in Fig. 8 [32]. Two LCD panels, the images of which are superimposed by a beam combiner, are used to display the left and right views. Field lenses placed close to the LCD serve to direct the illumination beams into the appropriate eye. For headtracking, the position of the light sources must be movable. The headtracking illumination system can, e.g., be implemented by monochrome CRTs displaying high-contrast camera images of the left and right halves of the observers's face [33]. Multiple user access is possible by using multiple independent illuminators.


      Fig. 8: Schematic view of a dual-LCD field-lens display.

A different solution developed at Dresden University uses a single panel and light source in connection with a prism mask (Fig. 9). Alternating pixel columns (RGB-tripels) are composed of corresponding columns of the left and right image. The prisms serve to deflect the rays of light to separated viewing zones.


      Fig. 9: Schematic view of a single-LCD field-lens display.

A field-lens based display with a movable (gaze-contolled) image plane is under development at Heinrich-Hertz-Institut Berlin [34]. As shown in Fig. 10, the stereo images are not projected onto a physical display screen, but appear in the form of aerial images floating in front of or behind a large Fresnel-type field-lens. The position of the image plane is controllable by motorized focus adjustments of the projection optics. A gaze tracker is used to sense the observer's momentary point of fixation, and the aerial image plane is moved to this position by corresponding focus adjustments. As the observer accommodates on the distance of the aerial-image plane, accommodation distance and convergence distance coincide like in natural vision. Additionally, the display concept encompasses a natural depth-of-focus effect by depth-selective spatial low-pass filtering of the projected images (depth-of-interest display concept [35]).


      Fig. 10: Components of a depth-of-interest display.


Reflection-based approach

This approach uses a retro-reflective screen for direction multiplexing. Retro-reflective means that the incident rays of light are reflected only into their original direction [22]. In a recent prototype [36], dual video projectors are mounted on a laterally movable stage (Fig. 11). The screen reflects the two images through a large half mirror to the observer's eyes. The system locates the current head position and adjusts the position of the projectors and the angle of the half-mirror accordingly.


      Fig. 11: Schematic view of the Xenotech display [36].

Occlusion-based approaches

These approaches have one thing in common: due to parallax-effects, parts of the image are hidden from the one eye but visible for the other eye. The technical solutions differ in the number of viewing slits (ranging from a dense grid to a single vertical slit), in presentation mode (time-sequential vs stationary) and in whether the opaque barriers are placed in front of or behind the image screen (parallax barrier vs parallax illumination techniques).

Basically, any of the lenticular display designs can also be implemented with parallax effects, replacing the raster of cylindrical lenslets by a raster of vertical slit openings (cf. Fig. 5 and 12). Parallax barriers are much easier to manufacture than lenticular screens (e.g. by printing or electro-optical methods). They can be moved electronically for headtracking and time sequential presentation (moving slit technique). Moreover, the dark barriers increase the maximal contrast under ambient illumination (black-matrix effect). The following paragraphs list some recent developments of parallax displays.

Barrier-grid displays: Sanyo has recently optimized the barrier-grid design for LCD-based direct-view displays [37]. One barrier is in front of the LCD and an additional one is placed between the LCD panel and the backlight case. The additional barrier has a reflective coating facing the backlight. It serves to exploit the light which would otherwise be blocked by the black matrix of the LCD panel. This way, lighting efficiency was improved by a factor of 1.4.


      Fig 12: Principle of a parallax-barrier direct-view 3-D display and grid vs pixel  arrangement for standard direct-view panels with vertical color-filter stripes.

As the rays of light pass through several adjacing slits, they produce a number of additional viewing windows for the left and right views. This can lead to the disturbing effect that the left eye perceives the right view and vice versa (i.e. the stereo-depth is inverted). Ideally, a headtracker should be used to switch the pixel-positions of the multiplexed left and right-eye images as soon as the observer enters a "wrong" viewing window [38]. In a prototype of NHK Labs [39], the barrier grid is generated by an LCD panel. This way, it can be limited to software-selected areas of the panel to create scalable "3-D windows". The grid is switched off in the 2-D area, making the full resolution of the imaging panel available.

Parallax-illumination displays: Dimension Technologies create the parallax-effect by a lattice of very thin vertical lines of light, placed at a distance behind a standard LC panel [40]. As indicated in Fig. 13, parallax causes each eye to perceive light passing through alternate image columns only.


      Fig. 13: Basic concept of a parallax illumination 3-D display

The illumination lattice is generated by means of a lenticular lens sheet which focuses the light from a small number of fluorescent lamps into a large number of light lines on the surface of a translucent diffusor. For headtracking, multiple sets of lamps and a large field lens are incorporated into the display. The position of the light source is changed by switching between sets of laterally displaced lamps. It is also possible to create two or more viewing zones for multiple observers by simultaneous operation of multiple light sources.

Two sets of blinking light sources, which generate two laterally displaced blinking lines behind each single column of the LCD, have been proposed to obtain full spatial resolution for both views. The sequential display of the left and right views is synchronized with the blinking light sources. An even more advanced version generates multiple sets of flashing light lines to create a stationary multiview display with look-around effect [41].

Moving-slit displays: With this technique, a single vertical slit opening serves to channel different perspective views to an array of adjacent viewing zones. The slit is generated by a fast-switching LCD mounted on the front of a CRT, and it passes the screen periodically at a rate of about 60 Hz. Image output of the CRT is synchronized with the slit position. Depending on the specific approach, the CRT screen displays either complete images or partial images composed of selected image columns of multiple views.

A monochrome display for up to 16 views has been developed at Cambridge University [42]. Complete views are presented on the screen for any given slit position and relayed to a circumscribed viewing window by means of three large spherical lenses (Fig. 14). As the slit traverses the screen, adjacent windows for the neighbouring views line up. The CRT requirements are very high (960 Hz frame rate for the 16-view display).


      Fig. 14: Optical set-up of a moving-slit 3-D display. Lenses 1 and 2  project the CRT image onto the field lens whichs in turn projects the slit aperture into  viewing space.

A display proposed at the University of Nagoya [43] uses a special CRT with separate electron guns for the different views. The electron guns have horizontally deflecting magnets to produce fanbeams, and these fanbeams perform only vertical scans. Thus, the information of individual image-columns (one per view) is written onto separate rectangular areas of the CRT, whilst each pixel is horizontally expanded to the width of one such rectangular area (Fig. 15). The observation slit moves in synchronism with the horizontal scanning signal of the input images.


      Fig. 15: Schematic view of a moving-slit 3-D display with a fanbeam CRT.

The so-called Holotron [44] is a moving-slit display without additional lenses in front of the CRT. Corresponding columns of different views are displayed side-by-side behind the slit aperture (Fig. 16). As the slit moves laterally, a new set of multiplexed image columns is displayed on the CRT. This way, a set of N views is displayed as a sequence of partial images, composed of N columns, during a single pass of the slit. The horizontal deflection of the electron beam is restricted to the momentary position of the partial-image area. To reduce the CRT requirements (sampling frequency, phosphor persistence), a multiple electron gun CRT in conjunction with a multiple moving-slit panel has been proposed.


      Fig. 16: Moving-slit approach of the Holotron.



Recent developments in information technology have stimulated a growing demand for mature 3-D displays. As a consequence, 3-D display technology has rapidly gained momentum in terms of research and commercial activities. On the one hand, these projects concentrate on the utilization of novel base technologies in conjunction with rather conventional 3-D concepts for the custom design of marketable equipment for special applications (aided viewing systems). On the other hand, there is a rising interest in new concepts for universal purpose systems that do not fall behind familiar 2-D displays in terms of image quality and viewing comfort (autostereoscopic displays). Overall, the research scene, especially in the field of autostereoscopic displays, is characterized by a large variety of competing concepts, each having particular advantages and flaws. Thus, there are still enormous challenges waiting for future R&D work: the ultimate 3-D display has yet to be invented.



      1 Wöpking, M 'Viewing comfort with  stereoscopic pictures: An experimental study on the subjective effects of disparity  magnitude and depth of focus.' Journal of the SID 1995, 3(3), 101-103

      2 Faris, S 'Novel 3-D stereoscopic imaging technology.' Proc. SPIE 2177, 1994,  180-195

      3 Pastoor, S 'Human factors of 3D imaging: Results of recent research at  Heinrich-Hertz-Institut Berlin'. Proc. 2nd Int. Display Workshop, Hamamatsu, 1995,  69-72

      4 Bos, P J 'Time sequential stereoscopic displays: The contribution of phosphor  persistence to the "ghost" image intensity.' Proc. 1991 ITE Annual Convention,  Tokyo, 1991, 603-606

      5 Fukai, K, Amafuji, H, Murata, Y 'Color and high resolution head-mounted display.' Proc.  SPIE 2177, 1994, 317-324

      6 Sperlich, T 'Welt im Helm - Die VR kommt ins Wohnzimmer.' c't 1995, (11),  200-208

      7 Bolas, M T, Lorimer, E R, McDowall, I E, Mead, R X 'Proliferation of counterbalanced,  CRT-based stereoscopic displays for virtual environment viewing and control.' Proc.  SPIE 2177, 1994, 325-334

      8 Omura, K, Shiwa, S, Kishino, F '3-D Display with accommodative compensation (3DDAC)  employing real-time gaze detection.' SID 96 Digest 1996, 889-892

      9 Honda, T 'Dynamic holographic 3D display using LCD.' Asia Display '95, 1995,  777-780

      10 Sato, K 'Development of 3-D TV by electro-holography.' Proc. 2nd Int. Display  Workshop , Hamamatsu, 1995, 73-76

      11 Yoshikawa, H 'Digital holographic signal processing.' TAO First Int.Symp. 1993 (Telecom. Advancement Org. of Japan, 2-31-19, Shiba, Minato-ku, Tokyo 105, Japan),  S-4-2-3 - 4-2-11

      12 Hashimoto, N, Morokawa, S 'Motion-picture holography using liquid-crystal television  spatial light modulators.' SID 95 Digest 1995, 847-850

      13 Benton, S A 'The second generation of the MIT holographic video system.' TAO  First Int.Symp. 1993 (Telecom. Advancement Org. of Japan, 2-31-19, Shiba, Minato-ku,  Tokyo 105, Japan), S-3-1-3 - S-3-1-6

      14 Yamada, H, Akiyama, K, Muraoka, K, Yamaguchi, Y 'The comparison of three kinds of  screens for a volume scanning type 3D display.' TAO First Int.Symp. 1993 (Telecom.  Advancement Org. of Japan, 2-31-19, Shiba, Minato-ku, Tokyo 105), S-5-3-3 - S-5-3-10

      15 Soltan, P, Trias, J, Dahlke, W, Lasher, M, McDonald, M 'Laser-based 3D volumetric  display system - Second generation.' In: Morgan, K et.al. (Eds.) Interactive Technology  and the New Paradigm for Healthcare, IOS Press and Ohmsha, 1995, 349-357

      16 McAllister, D (Ed.) Stereo computer graphics and other true 3D technologies .  Princeton University Press, Princeton, 1993

      17 'Imaging displays: Holographic 3-D images float in free space.' Laser Focus World 1995, (June), 22-23

      18 Jones, M W, Nordin, G P, Kulick, J H, Lindquist, R G, Kowel, S T 'A liquid crystal  display based implementation of a real-time ICVision holographic stereogram display.' Proc.  SPIE 2406 , 1995, 154-164

      19 Toda, T, Takahashi, S, Iwata, F '3D video system using Grating Image.' Proc. SPIE  2406, 1995, 191-198

      20 Schulze, E 'Synthesis of moving holographic stereograms with high-resolution spatial  light modulators.' Proc. SPIE 2406, 1995, 124-126

      21 Trayner, D, Orr, E 'Autostereoscopic display using holographic optical elements.' Proc.  SPIE, 1996

      22 Okoshi, T Three-Dimensional Imaging Techniques. Academic Press, New York,  1976

      23 McCormick, M 'Integral 3D imaging for broadcast.' Proc. 2nd Int. Display Workshop ,  Hamamatsu, 1995, 77-80

      24 Börner, R 'Autostereoscopic 3D-imaging by front and rear projection and on flat  panel displays.' Displays 1993, 14(1), 39-46

      25 Jewell, M R, Chamberlin, G R, Sheat, D E, Cochrane, P, McCartney, D J '3-D imaging  systems for video communication applications.' Proc. SPIE 2409, 1995, 4-10

      26 Uematsu, S, Hamasaki, J 'Clinical application of lenticular stereographic image  system.' Int. Symp. 3D Image Technology and Arts (Univ. of Tokyo, Seiken Symposium)  1992, Vol.8, 227-234

      27 Hentschke, S 'Personenadaptiver autostereoskoper Monitor (PAAS) - eine Option für  den Fernseher?' Fernseh- und Kinotechnik. 1996, 50(5), 242-248.

      28 Takemori, D, Kanatani, K, Kishimoto, S, Yoshii, S, Kanayama, H '3-D Display with  large double lenticular lens screens.' SID 95 Digest 1995, 55-58

      29 Omura, K, Shiwa, S, Kishino, F 'Development of lenticular stereoscopic display  systems: Multiple images for multiple viewers.' SID 95 Digest 1995, 761-763

      30 Montes, J D, Campoy, P 'A new three-dimensional visualization system based on  angular image differentiation.' Proc. SPIE 2409, 1995, 125-132

      31 Isono, H, Yasuda, M, Takemori, D, Kanayama, H, Yamada, C, Chiba, K 'Autostereoscopic  3-D TV display system with wide viewing angles.' Euro Display '93 (SID) Strasbourg,  1993, 407-410

      32 Ezra, D, Woodgate, G J, Omar, B A, Holliman, N S, Harrold, J, Shapiro, L S 'New  autostereoscopic display system.' Proc. SPIE 2409, 1995, 31-40

      33 Omori, S, Suzuki, J, Sakuma, S 'Stereoscopic display system using backlight  distribution.' SID 95 Digest 1995, 855-858

      34 Boerger, G, Pastoor, S Autostereoskopisches Bildwiedergabegerät mit natürlicher  Kopplung von Akkommodationsentfernung und Konvergenzentfernung und adaptiver  Schärfentiefe. Patent pending 25.09.1995, AZ 195 37.499.1

      35 Blohm, W, Beldie, I P, Schenke, K, Fazel, K, Pastoor, S 'Stereoscopic image  representation with synthetic depth of field.' to appear in Journal of the SID

      36 The Xenotech 3-D auto stereoscopic display system. Technical Information.  Xenotech Australia, Suite 1,41 Walters Drive, Osborne Park, Western Australia 6017

      37 Hamagishi, G, Sakata, M, Yamashita, A, Mashitani, K, Nakayama, E, Kishimoto, S,  Kanatani, K 'New stereoscopic LC displays without special glasses.' Asia Display '95 1995, 791-794

      38 Shiwa, S, Tetsutani, N, Akiyama, K, Ichinose, S, Komatsu, T 'Development of  direct-view 3D display for videophones using 15 inch LCD and lenticular sheet.' IEICE  Trans. Information and Systems 1994, Vol. E77-D(9), 940-948

      39 Isono, H, Yasuda, M, Sasazawa, H 'Autostereoscopic 3-D LCD display using  LCD-generated parallax barrier.' Japan Display '92 1992, 303-306

      40 Eichenlaub, J 'An autostereoscopic display with high brightness and power  efficiency.' Proc. SPIE 2177, 1994, 4-15

      41 Eichenlaub, J, Hollands, D, Hutchins, J 'A prototype flat panel hologram-like  display that produces multiple perspective views at full resolution.' Proc. SPIE 2409,  1995, 102-112

      42 Travis, A R L, Lang, S R, Moore, J R, Dodgson, N A 'Time-multiplexed  three-dimensional video display.' SID 95 Digest 1995, 851-854

      43 Hattori, T 'Electro-optical autostereoscopic displays using large cylindrical  lenses.' Proc. SPIE 1457, 1991, 283-289

      44 Sombrowsky, R Verfahren und Vorrichtung zur Erzeugung stereoskopischer  Darstellungen. German patent DE 42 28 111 C1 (25.8.1992)

email to