Devices that display images that appear to be three-dimensional, especially those featuring binocular disparity and motion parallax, are useful for a wide variety of applications including: medical imaging and procedures; entertainment, movies, and computer gaming; advertising and merchandising; communications and teleconferencing; information display and data manipulation; virtual exercise; virtual tours; molecular and genetic engineering; military and security applications; navigation and telerobotics; and product development, mechanical design, and industrial production.

Humans use several visual cues to recognize and interpret three-dimensionality in images. Monocular cues can be seen with just one eye. Monocular cues for three-dimensional images include: lighting and shading; linear perspective; the relative sizes of familiar objects of known size; objects in the foreground overlapping objects in the background (called "occlusion"); adjusting eye muscles to focus on an object at one distance while objects at other distances are out of focus (called "accommodation"); and objects moving relative to each other when one's head moves (called "motion parallax"). Binocular cues require two eyes. Binocular cues for three-dimensional images include: seeing an object or scene from slightly different perspectives in one's right and left eyes (called "binocular disparity" or "stereopsis"); and the intersection of the viewing axes from one's right and left eyes (called "convergence"). When a method of displaying three-dimensional images provides some of these visual cues, but not others, then the conflicting signals can cause eye strain and headaches for the viewer. 

The ultimate goal for methods of displaying three-dimensional images is to provide as many of these visual cues for three-dimensionality as possible while also: providing good image resolution and color; enabling large-scale displays; being viewable simultaneously by multiple viewers in different positions; not requiring special headgear; and being safe. This goal has not yet been achieved by current methods for displaying three-dimensional images. We now discuss twelve categories of methods for three-dimensional display, their limitations, and some examples of prior art that appear to use these methods. Since devices sometimes uses multiple methods, it is not always possible to neatly categorize examples of prior art into just one category, but the exercise and the categorization framework are nonetheless useful for structuring a review.

A long-standing method for displaying three-dimensional images involves glasses, or other headgear, that display slightly different views of an object or scene to a viewer's right and left eyes. This difference is called "binocular disparity". When the images that are seen in the right and left eyes are different perspectives of the same object or scene, as one would see when viewing the object or scene in the real world, then the brain interprets these two images synergistically as a single three-dimensional image. This is called "stereoscopic vision" or "stereopsis". 

There are three general ways in which glasses, or other headgear, can present different images to the two eyes using current technology. The first way involves lenses with different filters for the right and left eyes. For example, different color lenses (such as red vs. cyan) can each filter a different color in order to present different right and left views of an object to the right and left eyes. As another example, lenses with different polarizations (such as two perpendicularly-differing linear polarizations or two counter-rotational circular polarizations) can filter different image orientations in order to present different right and left views of an object to the right and left eyes. A second way involves glasses, or other headgear, with sequentially-alternating shutters on the right and left eyes. These sequentially-alternating shutters allow different right and left views to reach the right and left eyes in a time-sequential manner. A third way involves headgear with two independent image projectors, one for each eye, that independently display different right and left views to the right and left eyes.

Limitations of three-dimensional display using glasses or other headgear include: (1) inconvenience of glasses, or other headgear, for people who do not normally wear glasses and potential incompatibility with regular glasses for people who do normally wear glasses; (2) no motion parallax (at least without viewer head tracking, which addresses some limitations but creates others) and, as a result, multiple viewers see the same image from the same perspective regardless of their location or movement; and (3) conflict between accommodation and stereoscopic vision that can cause eye strain, headaches, and long-term adverse effects.

Due to the problems with three-dimensional display using glasses, or other headgear, identified above, there have been efforts to develop methods of three-dimensional display that do not require glasses or other headgear. Devices for displaying three-dimensional images with binocular disparity that do not require glasses or other headgear are called "autostereoscopic." One general category of autostereoscopic devices involves devices with stationary optical components that do not move in real time during imaging. (Such devices may have components that shift, or otherwise move, when they are switched between a two-dimensional display mode and a three-dimensional display mode, but if movement does not occur in real time during imaging then we classify them as having stationary optics.) Types of devices in this general category include: stationary volumetric displays; displays using stationary parallax barriers or lenticular arrays; stationary multi-angle sub-pixel display elements; stationary "fly's eye" lens or pin-hole arrays; and stationary rewritable holographic media. We now discuss each of these types of devices in greater detail.

Volumetric displays have one or more imaging surfaces that actually span a three-dimensional space. In this respect, volumetric displays create images that do not just appear to be three-dimensional, the images actually are three-dimensional. We define a stationary volumetric display as a volumetric display with image projection or light-emitting surfaces that do not move. One long-standing type of stationary volumetric display is a stationary curved projection surface, such as a cylindrical or hemispherical projection surface. Many planetariums use a dome-shaped projection surface for volumetric display. The audience sits under the dome while light beams representing stars and planets are projected onto the dome, creating the effect of a three-dimensional sky. Another type of stationary volumetric display consists of multiple layers (sometimes called "stacks") of light-reflecting projection surfaces with controllable opacity or with light-emitting arrays on transparent surfaces. These displays can be made from Polymer Dispersed Liquid Crystals (PDLCs), Liquid Crystal gel (LC-gel) elements, or arrays of Light Emitting Diodes (LEDs). For devices with light-reflecting projection surfaces, images are generally projected onto different layers in rapid succession, in synchronization with changes in the opacity of different layers. When these changes are sufficiently rapid, images on all layers appear simultaneously to the viewer due to persistence of vision. 

Some volumetric displays have a lens or an array of microlenses whose focal lengths can be changed in real time during image projection. The ability to change the lens focal length allows one to project images onto different distance surfaces for creating three-dimensional images. Different methods for changing the focal length of a microlens include: applying an electric potential to a polymeric or elastomeric lens; mechanically deforming a liquid lens sandwiched within a flexible casing; and changing the temperature of the lens. It is a judgment call whether to include devices with variable focal-length lenses (but projection surfaces that do not move) among stationary volumetric displays or whether to include them among moving volumetric displays that we will discuss later. For this review, we have chosen to focus on movement of the projection surface as the primary way to differentiate volumetric displays. Accordingly, we include volumetric displays with variable focal-length lenses and stationary projection surfaces within the general category of stationary volumetric displays.

The limitations of stationary volumetric displays include: (1) images tend to be ghost-like, with no opacity and no occlusion of foreground objects over background objects; (2) image resolution tends to be low and color variation is limited; (3) large amounts of data processing are often required; (4) for displays with stacks of display panels or three-dimensional arrays of light-emitting elements (such as LEDs), there is a dramatic increase in complexity, bulk, weight, and cost with increased display size; (5) for displays with stacks of display panels, there can be undesirable interference patterns as light passes through several panels; (6) for displays in a self-contained volume that is physically isolated from a viewer's hands, there is limited capability for touch-based interaction; and (7) for displays with arrays of light-emitting elements, dark boundaries between the light-emitting elements can create lines, graininess, and rough edges.

Three-dimensional image displays that use stationary parallax barriers or lenticular arrays have parallax barriers (light barriers) or lenticules (lenses) that do not move in real time during imaging. Some such devices may have layers or other components that move when shifting between two-dimensional and three-dimensional display modes, but this is not real time movement during imaging. Parallax barriers are structures that selectively block and transmit light from different portions of a light-emitting, or light-reflecting, surface in order to present the right and left eyes with different perspectives to create binocular disparity and stereopsis. For example, the display surface can show a composite image with vertical image stripes for right and left eye images and the parallax barrier can have vertical slits that direct the appropriate image stripes to reach the right and left eyes when the viewer is located within a restricted viewing location. Generally, if the viewer moves outside the restricted viewing location, then the viewer sees undesirable "pseudoscopic" images with reversed depth, double images, or black lines. These pseudoscopic images can cause eye strain and headaches. 

Having some distance between the parallax barrier and a light-emitting, or light-reflecting, surface is required in order for the parallax barrier to direct light rays along different angles to the right and left eyes. However, this distance causes many of the limitations of the parallax barrier method. For example, this distance restricts the proper viewing location within which the viewer must be located in order to avoid pseudoscopic images. This distance is also why parallax barriers do not work well, if at all, for simultaneous viewing by multiple viewers and why motion parallax is limited with parallax barriers.

Lenticules are lenses, generally configured in a lens-repeating array, that selectively steer different portions of an image from a light-emitting, or light-reflecting, surface to the right and left eyes in order to create binocular disparity, stereopsis, and motion parallax. Lenticular arrays may be configured in a single layer or in multiple parallel layers. The most common lenticule configurations are arrays of vertical plano-convex columns, bi-convex columns, or semi-cylindrical columns. Vertical lenticular columns create some motion parallax when a viewer moves their head from side to side, but not when they move their head up and down. Lenticular lens columns may also be arranged horizontally, allowing motion parallax when a viewer moves their head up and down, but not when they move their head from side to side. 

Motion parallax from lenticular arrays is generally limited to a modest number of sequential views. Three-dimensional image displays using lenticular arrays display only a limited number of different images as a viewer's head moves. The changing images may appear to come from the same location, but they actually come from different locations associated with each lenticule. This is called spatial demultiplexing. When spatial demultiplexing is accomplished using lenticules, then the number of alternative views (the range of motion parallax) is limited by the space constraints of the lenticule. If the lenticule is large, then the image has low resolution. If the lenticular is small, then the number of different views is quite limited. With current technology, it is rare to have a lenticule-based display that offers more than ten alternative perspectives as a viewer moves their head. This range can be expanded somewhat with additional technology such as head tracking, which we will discuss later in a section on devices with moving optics, but for stationary lenticular arrays the number of different views for motion parallax is generally quite limited.

Due to the spatial constraints of displaying multiple views from different places (eg. strips) so that they are seen coming from the same location (eg. same strip), there is a loss of image resolution in three-dimensional display devices that use lenticular arrays. For this reason, some devices are designed to be switched from a two-dimensional display mode with higher resolution to a three-dimensional display mode with lower resolution. This can be done by shifting one or more lenticular arrays relative to each other or relative to an image display surface. In a first configuration, the two layers perform demultiplexing for three-dimensional display. In a second configuration, the two layers do not perform demultiplexing for two-dimensional display. When this shifting does not occur in real time during imaging, then we classify such mode-shifting devices as having stationary optics.

The limitations of three-dimensional displays that use stationary parallax barriers or lenticular arrays include: (1) the viewing zone is restricted and outside this restricted zone a viewer sees pseudoscopic images (with depth reversal, double images, and black bands); (2) there is a tradeoff between low resolution and limited range of motion parallax (generally less than twelve different perspectives) due to the constraints of spatial demultiplexing; (3) for displays with vertical lenticules or vertical parallax barriers, there is little or no vertical motion parallax (with up and down head motion); (4) such devices are generally restricted to one viewer; (5) lenticular arrays and active parallax barriers, such as Liquid Crystal Display (LCD) shutters, can be expensive and fragile; (6) for displays with parallax barriers, the image can be dim because the barriers block a significant amount of the image light; (7) conflict between accommodation and stereoscopic vision can cause eye strain, headaches, and long-term adverse effects; and (8) boundaries between light-emitting elements can create dark lines, graininess, and rough edges. 

Examples in the prior art that appear to use stationary parallax barriers or lenticular arrays include the following U.S. patents -- Nos. 5,550,676 (Ohe et al., 1996), 5,790,086 (Zelitt, 1998), 5,982,342 (Iwata et al., 1999), 6,064,424 (van Berkel et al., 2000), 6,201,565 (Balogh, 2001), 6,547,400 (Yokoyama, 2003), 6,606,078 (Son et al., 2003), 6,795,241 (Holzbach, 2004), 6,843,564 (Putilin et al., 2005), 6,876,495 (Street, 2005), 7,084,841 (Balogh, 2006), 7,250,990 (Sung et al., 2007), 7,265,902 (Lee et al., 2007), 7,268,943 (Lee, 2007), 7,342,721 (Lukyanitsa, 2008), 7,382,425 (Sung et al., 2008), 7,400,447 (Sudo et al., 2008), 7,423,796 (Woodgate et al., 2008), 7,425,951 (Fukushima et al., 2008), 7,426,068 (Woodgate et al., 2008), 7,471,352 (Woodgate et al., 2008), 7,492,513 (Fridman et al., 2009), and 7,506,984 (Saishu et al., 2009).

Examples in the prior art that appear to use stationary parallax barriers or lenticular arrays also include the following U.S. patent applications -- Nos. 20030206343 (Morishima et al., 2003), 20040150583 (Fukushima et al., 2004), 20050041162 (Lee et al., 2005), 20050073577 (Sudo et al., 2005), 20060176541 (Woodgate et al., 2006), 20060279680 (Karman et al., 2006), 20070035829 (Woodgate et al., 2007), 20070058127 (Mather et al., 2007), 20070097019 (Wynne et al., 2007), 20070222915 (Niioka et al., 2007), 20080150936 (Karman, 2008), 20080204873 (Daniell, 2008), 20080231690 (Woodgate et al., 2008), 20080273242 (Woodgate et al., 2008), 20080297670 (Tzschoppe et al., 2008), 20080309663 (Fukushima et al., 2008), 20090002262 (Fukushima et al., 2009), 20090046037 (Whitehead et al., 2009), 20090079728 (Sugita et al., 2009), 20090079733 (Fukushima et al., 2009), 20090096726 (Uehara et al., 2009), and 20090096943 (Uehara et al., 2009).

Another method of displaying images that appear to be three-dimensional involves stationary multi-angle sub-pixel display elements. Stationary multi-angle sub-pixel display elements are relatively complex. They have "pixels within pixels," sometimes called "sub-pixels," wherein each sub-pixel has a light-channeling structure that directs light rays from the sub-pixel toward the viewer at a different exit angle. This allows one to display different light content (eg. different color and intensity) from different angles from the same spot (eg. the same pixel) on a display surface. With an array of multi-angle sub-pixel display elements, one can create different views of the same object as seen from different locations, thereby creating images that appear to be three-dimensional with binocular disparity and motion parallax. 

A stationary multi-angle sub-pixel display element has a number of discrete fixed-location light channels at the sub-pixel level, each of which channels light in a different direction. For example, a stationary multi-angle sub-pixel display element may be a dome-shaped structure that contains an array of fiber optics that each radiate out toward the perimeter of the dome at different angles. In another example, a stationery multi-angle sub-pixel display element may be a stationary concave structure with a central Light Emitting Diode (LED) and multiple lenses that direct light from the LED into different exit angles. In another example, a display element of this type may have a single micro lens and multiple LEDs, wherein light rays from the multiple LEDs pass through the same lens in different trajectories and exit the lens at different angles. 

In concept, using stationary multi-angle sub-pixel display elements can be a very powerful method for producing images that appear to be three-dimensional. However, at least with present-day technology, these structures have significant limitations that constrain image resolution and motion parallax. For example, due to the discrete and stationary nature of the sub-pixel elements, and the space constraints involved in structures as small as one pixel, there are limitations on how many different sub-pixel elements one can pack into one display element. This limits the number of discrete exit angles that one can achieve with such structures. For example, how many individual LEDs, optical fibers, microlenses, or other light-channeling sub-pixels at different angles can one fit into a space the size on one pixel? At least with present-day technology, the answer is unlikely to be sufficiently large to provide a high-resolution image with a significant range of motion parallax.

The limitations of multi-angle sub-pixel display elements include: (1) low image resolution, limited number of perspectives, and limited viewing range due to spatial constraints on how many discrete, stationary sub-pixel elements one can fit into a small space the size of one pixel; (2) due to the complexity of the microstructures required to direct light rays along different angles at the pixel-specific level and due to the large numbers of microstructures required in a display array, there is a dramatic increase in display complexity, bulk, and cost with increased display size; (3) conflict between accommodation and stereoscopic vision can cause eye strain, headaches, and long-term adverse effects; and (4) boundaries between light-emitting elements can create dark lines, graininess, and rough edges. 

Examples in the prior art that appear to use multi-angle sub-pixel display elements include U.S. Patent Nos. 5,132,839 (Travis, 1992), 5,689,321 (Kochi, 1997), 6,128,132 (Wieland et al., 2000), 6,344,837 (Gelsey, 2002), 6,736,512 (Balogh, 2004), and 7,446,733 (Hirimai, 2008), and U.S. Patent Application No. 20050285936 (Redert et al., 2005).

Another method of displaying images that appear to be three-dimensional involves the use of "fly's eye" lens or pin-hole arrays. "Fly's eye" lens arrays have an array of semi-spherical lenses. When fly's eye lenses are used to take pictures, the process is called "integral photography." In some respects, fly's eye lenses are semi-spherical versions of the linear columnar lenses that are used in common lenticular arrays. Pin-hole arrays have an array of point openings through which an image is viewed. In some respects, pin-hole arrays are semi-spherical versions of the linear slits that are used in common parallax barriers. Fly's eye lens and pin-hole arrays can provide some motion parallax in both vertical and horizontal directions, but have limitations in terms of low image resolution and limited image brightness.

The limitations of fly's eye lens and pin-hole arrays include: (1) images created using pinhole arrays tend to be dim and have low resolution; (2) pinhole and fly's eye lens arrays do not capture and display information from the entire surface of an object; (3) fly's eye lenses tend to be expensive to make; (4) conflict between accommodation and stereoscopic vision can cause eye strain, headaches, and long-term adverse effects; and (5) boundaries between light-emitting elements can create dark lines, graininess, and rough edges. 

Another method of displaying images that appear to be three-dimensional involves rewritable holographic media. Holography involves recording and reconstructing the amplitude and phase distributions of an interference pattern of intersecting light beams. The light interference pattern is generally created by the intersection of two beams of coherent light: a signal beam that is reflected off (or passed through) an object and a reference beam that comes from the same source. When the interference pattern is recreated and viewed by an observer, it appears as a three-dimensional object that can be seen from multiple perspectives. 

The ability to create non-rewritable holograms has existed for several decades. Non-rewritable holograms create: a limited range of motion parallax for seeing different perspectives of a static object as a viewer moves their head; or a limited series of changing images from the same surface that are seen as a viewer moves their head. Until recently, progress toward holographic animation with motion parallax for animated content (that changes independently of viewer motion) with rewritable holographic media has been limited. However, recently there has been a breakthrough in the use of rewritable holographic media that was achieved by researchers at the University of Arizona (Savas¸ Tay et al., "An Updatable Holographic Three-Dimensional Display," Nature, 451, Feb 7, 2008). This breakthrough involves photosensitive media in which holograms can be encoded, erased, and then re-encoded with sufficient speed such that holographic animation (also called "holographic video") is observed. Although this line of research has considerable potential, it is still in an early stage and quite limited with respect to image size, resolution, color, and speed. 

The limitations of stationary rewritable holographic media include: (1) image size and resolution are very limited, at least with current technology; (2) the refresh rate (frame speed) is very limited, at least with current technology; (3) there is limited color variation; (4) images tend to be ghost-like; (5) speckle interference patterns and undesirable quantum interactions can occur; (6) this technique can require nearly-darkroom conditions to take pictures for imaging; and (7) there can be cost and safety issues associated with coherent (eg. laser) light.

Another category of devices for displaying three-dimensional images involves displays that have optical components that shift linearly (such as side-to-side or in-and-out) on an image-wide level. In this review, displays with image-wide linearly-shifting optics are distinguished from: displays with optical components that shift independently at the level of individual pixels; and displays with optical components, at any level, that rotate or move in some other non-linear manner. 

One type of display with image-wide linearly-shifting optics is a parallel-shifting volumetric display, wherein one or more projection surfaces shift in parallel through a display volume. Due to persistence of vision, this movement creates a sequence of parallel two-dimensional images in space that, together, comprise a three-dimensional image. Limitations of parallel-shifting volumetric displays include: (1) images tend to be ghost-like, with no opacity or occlusion of foreground objects over background objects; (2) resolution tends to be low and color variation is limited; (3) large amounts of data processing are often required; (4) for displays with moving display surfaces, there is complexity, inertial stress, and mechanical wear and tear associated with larger displays; (5) for displays in a self-contained volume that is physically isolated from a viewer's hands, there is no capability for touch-based interaction; (6) for displays with arrays of active (variable-length) microlenses, larger size displays can be complex and expensive; (7) for displays with arrays of active (variable-length) microlenses, there are constraints on how rapidly the microlenses adjust, especially those with fluid components; and (8) for displays with arrays of light-emitting elements, dark boundaries between the light elements can create lines, graininess, and rough edges.

Another type of display with image-wide linearly-shifting optics is a spatially-demultiplexing device with one or more shifting layers of light-emitting, light-reflecting, light-blocking, or light-refracting layers that include one or more parallax barriers or lenticular arrays. One or more of these layers shift relative to each other, in a linear side-to-side or in-and-out manner, often in parallel planes, at an image-wide level. In display systems with head-tracking mechanisms, the shifting motion of the optical layers can be coordinated with movement of a viewer's head to extend the viewing range in which proper autostereoscopic images are seen and to reduce the chances of pseudoscopic images. In display systems wherein one layer has a scanning hole or lens, the shifting motion of the optical layers can extend the number of views in spatial demultiplexing. 

Limitations of displays with image-wide linearly-shifting parallax or lenticular layers include: (1) for displays with head tracking in which optical layers shift in response to viewer head movement, the viewing zone can be expanded but it is still limited, head tracking can be cumbersome, and head tracking does not work well with multiple viewers; (2) there is a still a tradeoff between low horizontal resolution and limited range of motion parallax due to the constraints of spatial demultiplexing; (3) for displays with vertical lenticules or vertical parallax barriers, there is little or no vertical motion parallax (with up and down head motion); (4) mechanical wear and tear associated with real-time moving optical layers, especially for larger displays; (5) lenticular arrays and active parallax barriers, such as Liquid Crystal Display (LCD) shutters, can be expensive and fragile; (6) for displays with parallax barriers, the image can be dim because the barriers block a significant portion of the image light; (7) conflict between accommodation and stereoscopic vision can cause eye strain, headaches, and long-term adverse effects; and (8) boundaries between light elements can create dark lines, graininess, and rough edges. 

Examples in the prior art that appear to have image-wide linearly-shifting parallax or lenticular layers include U.S. Patent Nos. 4,740,073 (Meacham, 1988), 5,300,942 (Dolgoff, 1994), 5,311,220 (Eichenlaub, 1994), 5,602,679 (Dolgoff et al., 1997), 5,825,541 (Imai, 1998), 5,872,590 (Aritake et al., 1999), 5,900,982 (Dolgoff et al., 1999), 6,014,164 (Woodgate et al., 2000), 6,061,083 (Aritake et al., 2000), 6,791,512 (Shimada, 2004), 6,798,390 (Sudo et al., 2004), 7,030,903 (Sudo, 2006), 7,123,287 (Surman, 2006), 7,283,308 (Cossairt et al., 2007), 7,375,885 (Ijzerman et al., 2008), 7,432,892 (Lee et al., 2008), 7,450,188 (Schwerdtner, 2008), and 7,532,225 (Fukushima et al., 2009), and U.S. Patent Application Nos. 20030025995 (Redert et al., 2003), 20030058209 (Balogh, 2003), 20030076423 (Dolgoff, 2003), 20040178969 (Zhang et al., 2004), 20050219693 (Hartkop et al., 2005), 20050264560 (Hartkop et al., 2005), 20050280894 (Hartkop et al., 2005), 20060109202 (Alden, 2006), 20080117233 (Mather et al., 2008), 20080204873 (Daniell, 2008), 20090040753 (Matsumoto et al., 2009), and 20090052027 (Yamada et al., 2009).

Another category of devices for displaying three-dimensional images involves displays with optical components that rotate (or move in some other non-linear manner) on an image-wide level. 

Rotating volumetric displays generally create three-dimensional images by projecting a series of two-dimensional images onto a rotating surface. When the surface rotates sufficiently rapidly, then this series of two dimensional images is perceived as being simultaneous by a viewer due to persistence of vision. When this series of two-dimensional images comprise views of the same object from different perspectives and these different views are coordinated with the angular movement of the rotating surface, then this creates the perception of a three-dimensional object that can be viewed from different perspectives as one or more viewers move around the display (an effect that is called "angular motion parallax" or "theta parallax"). 

The rotating surface in such devices may diffuse, reflect, or refract light from the image projected onto it. The shape of the rotating surface may be a disk, square, helix, wedge, pyramid, or some other shape. The rotating surface is generally enclosed to protect the viewer from being harmed by contact with the rapidly rotating surface. It can be challenge to keep the image in focus as the surface rotates. Methods to keep the image in focus as the surface rotates include: a moving projector or reflector that moves in synchronization with the rotating surface; multiple projectors that project images in sequence around the rotating surface; and lenses with variable focal lengths that can be changed in real time. Another type of rotating volumetric display involves light-emitting elements on the rotating surface itself, but connections to light-emitting members on the rotating surface are complex and those light-emitting elements are subject to considerable stress from inertial forces at high rotation rates. 

The limitations of rotating volumetric displays include: (1) images tend to be ghost-like, with no opacity or occlusion of foreground objects over background objects; (2) it is difficult to have larger displays due to the mass, inertia, and structural stress of large rapidly-spinning objects; (3) there are issues with the complexity, mechanical wear, and noise of rotary bearings and other moving parts; (4) for displays in which the angle between a screen and projection beam sometimes becomes small during portions of the rotation, the image quality is decreased during such times; and (5) for displays housed in a self-contained volume that is physically isolated from a viewer's hands, there is limited capability for touch-based interaction. 

Examples in the prior art that appear to have rotating volumetric displays include U.S. Patent Nos. 4,160,973 (Berlin, 1979), 5,148,310 (Batchko, 1992), 6,816,158 (Lemelson et al., 2004), 7,023,466 (Favalora et al., 2006), 7,277,226 (Cossairt et al., 2007), 7,364,300 (Favalora et al., 2008), 7,490,941 (Mintz et al., 2009), and 7,525,541 (Chun et al., 2009), and U.S. Patent Application Nos. 20050152156 (Favalora et al., 2005), 20050180007 (Cossairt et al., 2005), and 20070242237 (Thomas, 2007).

Another type of three-dimensional display using image-wide rotating (or other non-linearly-moving) optical components is a display with a image-wide rotating (or otherwise non-linearly-moving) lens, light barrier, or mirror. For example, a spinning optical lens with angularly-varying thickness or ridges can be placed in front of an imaging surface. This spinning lens can change the focal distance of elements of the projected image in a rapid, cyclical manner to create three-dimensional effects. As another example, an image may be projected through radial slits in a spinning disk. In another example, a beam of (coherent) light may be reflected off a spinning polygonal mirror onto a diffuser. When changes in the focal distances and/or exit angles of light rays passing through, or reflecting off, an image-wide rotating (or otherwise non-linearly-moving) lens, light barrier, or mirror are coordinated with changes in contents of those light rays, then some three-dimensional effects can be achieved. However, at least with present-day technology, these three-dimensional effects are limited because image-wide rotating members do not provide independent control of exit angles at the level of individual pixels.

The limitations of three-dimensional displays with image-wide rotating (or other non-linearly-moving) optical components include: (1) it can be difficult to achieve motion parallax and binocular disparity with a single image-wide rotating optical member, especially using non-coherent light; (2) the display size can be limited by the inertial forces and stresses of large spinning objects; (3) there are cost, complexity, wear, and noise issues associated with moving components; (4) conflict between accommodation and stereoscopic vision can cause eye strain, headaches, and long-term adverse effects; and (5) rotating mirrors tend to work mainly with coherent light, which can have associated cost and safety issues.

Examples in the prior art that appear to have three-dimensional displays with image-wide rotating (or other non-linearly-moving) optical components include U.S. Patent Nos. 3,199,116 (Ross, 1965), 3,602,572 (Norris, 1971), 5,111,313 (Shires, 1992), 5,694,235 (Kajiki, 1997), 5,704,061 (Anderson, 1997), 6,061,489 (Ezra et al., 2000), 6,115,059 (Son et al., 2000), 6,533,420 (Eichenlaub, 2003), 6,819,489 (Harris, 2004), 6,999,071 (Balogh, 2006), 7,036,935 (Shpizel, 2006), 7,113,158 (Fujiwara et al., 2006), 7,182,463 (Conner et al., 2007), 7,300,157 (Conner et al., 2007), 7,492,523 (Dolgoff, 2009), and 7,513,623 (Thomas, 2009), and U.S. Patent Application Nos. 20020084951 (McCoy, 2002), 20030067421 (Sullivan, 2003), 20050248972 (Kondo et al., 2005), 20050270645 (Cossairt et al., 2005), 20060023065 (Alden, 2006), 20060109200 (Alden, 2006), 20060203208 (Thielman et al., 2006), and 20060244918 (Cossairt et al., 2006).

Another category of devices for displaying three-dimensional images involves displays with optical components that shift linearly (e.g. side-to-side or in-and-out) or tilt (e.g. to one side or the other) at the level of individual pixels in real time during imaging. 

One example of such optical components is a variable focal-length microlens. Variable focal-length microlenses are microscale lenses whose focal lengths can be changed in real time during imaging. Such microlenses are often called "active" or "dynamic." Different methods for changing the focal length of a microlens include: applying an electric potential to a polymeric or elastomeric lens; mechanically deforming a liquid lens sandwiched within a flexible casing; and changing the temperature of the lens. We have already discussed variable-focal-length lenses in the context of volumetric displays where they are used to focus images on different two-dimensional layers or on a rotating projection surface. This present category of devices includes displays that use variable focal-length microlenses, but are not volumetric. An array of variable focal-length microlenses may be used in combination with multiple parallel lenticule layers. Changing the focal length of a microlens can focus light rays from a pixel on a different lenticular layer, changing the exit angle of light rays from a given pixel and creating images that appear to be three-dimensional.

The limitations of non-volumetric three-dimensional displays using pixel-specific variable focal-length microlenses include: (1) for displays with arrays of active (variable-length) microlenses, larger size displays can be complex and expensive; (2) for displays with arrays of active (variable-length) microlenses, there are constraints on how rapidly the microlenses adjust in real time, especially those with fluid components; and (3) conflict between accommodation and stereoscopic vision can cause eye strain, headaches, and long-term adverse effects. 

Examples in the prior art that appear to have non-volumetric three-dimensional displays using pixel-specific variable-focal-length microlenses include U.S. Patent Nos. 5,465,175 (Woodgate et al., 1995), 5,493,427 (Nomura et al., 1996), 5,581,378 (Kulick et al., 1996), 5,790,086 (Zelitt, 1998), 5,801,761 (Tibor, 1998), 5,986,811 (Wohlstadter, 1999), 6,014,259 (Wohlstadter, 2000), 6,437,919 (Brown et al., 2002), 6,437,920 (Wohlstadter, 2002), 6,665,108 (Brown et al., 2003), 6,714,174 (Suyama et al., 2004), 6,755,534 (Veligdan et al., 2004), 6,831,678 (Travis, 2004), 6,909,555 (Wohlstadter, 2005), 7,046,447 (Raber, 2006), 7,106,519 (Aizenberg et al., 2006), 7,167,313 (Wohlstadter, 2007), 7,204,593 (Kubota et al., 2007), 7,297,474 (Aizenberg et al., 2007), 7,327,389 (Horimai et al., 2008), 7,336,244 (Suyama et al., 2008), and 7,480,099 (Raber, 2009).

Examples in the prior art that appear to have non-volumetric three-dimensional displays using pixel-specific variable-focal-length microlenses also include patent applications such as -- U.S. Patent Application Nos. 20040141237 (Wohlstadter, 2004), 20040212550 (He, 2004), 20050111100 (Mather et al., 2005), 20050231810 (Wohlstadter, 2005), 20060158729 (Vissenberg et al., 2006), 20070058258 (Mather et al., 2007), 20070165013 (Goulanian et al., 2007), 20070242237 (Thomas, 2007), 20080007511 (Tsuboi et al., 2008), 20080117289 (Schowengerdt et al., 2008), 20080266387 (Krijn et al., 2008), 20090021824 (Ijzerman et al., 2009), 20090033812 (Ijzerman et al., 2009), 20090052049 (Batchko et al., 2009), and 20090052164 (Kashiwagi et al., 2009).

Another example of three-dimensional displays that use pixel-specific, linearly-shifting or tilting optics are micromirror arrays with adjustable-angle mirrors at the level of individual pixels. Micromirror arrays are generally created and controlled using MEMS (Micro Electro Mechanical Systems). The angle of each mirror is adjusted in real time, during imaging, to change the exit angle for light rays exiting each pixel over time. When the changing angles of light rays exiting specific pixel elements are coordinated with changes in the contents of those light rays (e.g. color and intensity), then this can create images that appear to be three-dimensional. 

Micromirror arrays are often used with coherent light, such as the light from lasers, because coherent light can be targeted onto, and bounced off, moving mirrors in a much more precise manner than is possible with incoherent light. In an example, an array of directed coherent light beams bouncing off a micromirror array can be intersected within a volume of translucent material to create changing holographic images.

The limitations of three-dimensional displays using moving micromirror arrays at the pixel-specific level include: (1) they require a large number of Spatial Light Modulators (SLMs), so image size and resolution are limited, at least with current technology; (2) there can be cost and safety issues associated with coherent (eg. laser) light; (3) they require large amounts of data processing, especially for interference fringe patterns; (4) they feature limited color variation; (5) they produce ghost-like images with no opacity and limited interposition; (6) resulting images may have speckle patterns associated with lasers; (7) generally low utilization efficiency of diffracted light for the space modulator when forming interference fringe patterns for real-time hologram animation; (8) mechanical limitations of moving parts; and (9) difficulties of dealing with quantum interactions.

Among the many examples in the prior art that appear to have three-dimensional displays using moving micromirror arrays at the pixel-specific level are U.S. Patent Nos. 6,259,450 (Chiabrera et al., 2001), 6,329,963 (Chiabrera et al., 2001), 6,956,687 (Moon et al., 2005), 7,261,417 (Cho et al., 2007), and 7,505,646 (Katou et al., 2009), and U.S. Patent Application Nos. 20040252187 (Alden, 2004) and 20090040294 (Smalley et al., 2009).