Volumetric 3D Displays

A volumetric display device permits the generation, absorption or scattering of visible radiation
from a set of localized and specified regions within a physical volume. In some cases a volumetric
system may support the controlled anisotropic propagation of this radiation from each localized
region.
The first sentence characterizes the traditional form of volumetric system in which an image is
formed by controlling the emission or scattering of ambient light from a set of image elements
(voxels) located at appropriate positions within a transparent 3D image space. This basic
functionality is then extended by taking into account recent developments aimed at controlling
the directions in which light can propagate from each activated voxel site. In combination, these
two facets of the volumetric approach suggest the formation of a so called light engine – offering
the spatial, temporal, and directional control of light output from an image space.
Volumetric images comprise a set of voxels located within a 3D volume – a ‘‘physical’’ image
space. Consequently, such images are inherently 3D and support in a natural manner the majority
of pictorial cues to depth, along with binocular and motion parallax. In addition, the
accommodation/convergence breakdown characteristic of stereoscopic systems is avoided.
The majority of volumetric displays developed to date are unable to support image opacity and
so are characterized by translucent image formation. This is not an inherent characteristic of the
volumetric paradigm and several traditional and emerging embodiments enable opaque images
to be generated. In the case of certain potential applications in which, for example, there is a
need to view the form and location of entities such as cancerous tumors relative to surrounding
tissue and structures, image translucency can be advantageous. Unfortunately, when
photographed using conventional techniques only pictorial cues are captured and when
compounded with the translucent nature of an image scene, the results are most unsatisfactory
(see >Fig. 1). This contrasts with the direct viewing scenario where the eyes naturally converge
and focus on the features of interest.
A volumetric display should exhibit uniformity in voxel placement. Thus an image space should
provide a homogeneous array of possible voxel locations so that if, for example, an image
component is subjected to a translation operation, the number and spatial distribution of voxels
from which it is formed should remain unaltered. Although an image space may offer a large
number of possible voxel locations (the voxel location capacity (NL)), within a single image update
period the system architecture may impose a limit on the total number that may be activated (the
voxel activation capacity (NA)), and so contribute to the visible image (clearly, in the extreme case,
enabling all possible voxels to be activated simply gives rise to a rather large light source). In this
context, an image space should exhibit predictability, thereby enabling any combination of NA
voxels to be selected from the available pool of voxel locations. Furthermore, a display should
exhibit invariance in voxel attributes, such that all voxels have identical dimensions and that each
responds to the activation stimulus in the same manner.
Fig. 1: Volumetric images support a broad range of pictorial cues to depth together with the
oculomotor and parallax cues. However, when these images are photographed, only the pictorial
cues are recorded – the others are lost. While a sense of true three-dimensionality can be
portrayed when very simple images are photographed, the results of photographing more
complex images are disappointing. Image (a) shows a simple stick figure that is able to walk and
run around the display volume in a natural manner. Image (b) shows a complex molecular
structure – but the true three-dimensionality of this structure is not evident in the photograph.
Both images are depicted on the Cathode Ray Sphere

The notions of uniformity, predictability, and invariance are summarized in >Fig. 2, and as
indicated in this illustration any deficiencies in these characteristics are likely to result in the
quality of an image component varying with its position and orientation in the image space. The
volumetric technologies developed to date generally exhibit nonoptimal characteristics in at least
one of these areas.
In principle, volumetric displays are able to provide considerable freedom in viewing location. The
most restricted case occurs when only a single ‘‘window’’ onto the image space is supported.
More usually, systems support a cylindrical or hemispherical form of image space and allow ‘‘all-
round’’ viewing. These are increasingly (and perhaps unfortunately) referred to as ‘‘3600”
displays.’’ The least restrictive technologies enable images to be viewed from any location about
a hemispherical image space. Unfortunately, in practice, components that form the image space
may inadvertently cause light emanating from activated voxels to travel in a non-rectilinear
manner, and opaque components may occlude the passage of light in some directions. In
addition, emerging light may be subjected to significant refraction at the image space boundary.
Such effects compromise perceived image quality and lead to preferential viewing directions,
where images appear at their best when created within a certain region of an image space and
when viewed from a certain orientation.

Fig. 2: An image space should provide a uniform and predictable tableau such that image quality
does not vary with image location or with viewpoint.
As indicated in >Fig. 2, it is critical that a volumetric display architecture supports not only a
uniform and predictable 3D tableau with respect to image formation, but that it is also optically
neutral thereby supporting the rectilinear propagation of light and minimizing the uncontrolled
anisotropic passage of light through the image space.
Additionally, the requirement for minimum refraction at the image space boundary suggests that
an image space comprising a solid or liquid medium should be avoided.
A volumetric display may be considered in terms of three highly integrated subsystems (see >Fig.
3). These relate to the methods used for the formation of the image space (the primary image
depiction subsystem), the underlying physical techniques used for the formation of visible voxels
(the image element generation subsystem), and the method employed for driving the voxel
generation subsystem – thereby effecting changes in visible voxel attributes (the image element
activation subsystem).
These subsystems are assumed to operate so as to support a 3D display tableau referred to as the
‘‘primary image space.’’
Techniques used for the formation of a physical volumetric image space prevent the insertion of
physical interaction tools into the volume, thus, for example, precluding the possibility of making
‘‘contact’’ with an image via a haptic interaction probe. Consequently, in >Fig. 3 provision is made
for the incorporation of a ‘‘reimaging/projection’’ subsystem that comprises optical components
able to project the physical image space into an ‘‘ethereal space’’ (referred to here as the
‘‘secondary image space’’).

Fig. 3: It is convenient to consider volumetric displays as comprising either three or four

subsystems. Traditional approaches have used only the lower three subsystems. However, the
reimaging/ projection subsystem provides the opportunity to cast the volumetric image into free
space, thereby forming an ‘‘ethereal’’ image space in which image components and physical
interaction tools may coexist.
a) Swept-Volume Techniques
In this scenario, the image space is formed by means of the rapid cyclic movement of a ‘‘screen’’
which is assumed to comprise a ‘‘surface of emission’’ (SOE) and any associated supporting
structure. In turn, the SOE is " A planar or curved surface which cyclically sweeps out an image
space and upon which voxels are generated. It may be passive (e.g., a light-scattering surface) or
active (e.g., a matrix of individually addressable light emitting components). In the case of swept-
volume displays, the SOE constitutes the voxel generation subsystem.

Fig. 4: Translational motion is used in conjunction with a planar screen. In the case of rotational
motion, the planar and helical forms of surface of emission (SOE) are the most likely
configurations. Other forms of screen may be used (such as the Archimedes spiral) but these are
less efficient approaches as only a portion of the region swept out by the screen can be used for
the image space.

Fig. 5: Exemplar approaches that may be adopted in the implementation of the image element
generation and activation subsystems for a swept-volume volumetric display
The screen may move with either rotational or translational motion, and in >Fig. 4, possible SOE
geometries are summarized.>Figure 5 indicates opportunities for implementing the voxel
generation and voxel activation subsystems.
Several exemplar scenarios are outlined below:
i) Translational Motion
Here a planar screen rapidly moves back and forth, thereby sweeping out an image space. The
3D image data set is ordered into a set of parallel slices, each of which is in turn output to the SOE
as it passes through the appropriate position in space. To avoid image flicker, the SOE must sweep
out the image space at a frequency equal to or in excess of the critical flicker frequency (CFF).
Systems employing translational motion have usually supported one or two viewing windows,
located at one or both ends of the screen’s cycle of movement. An ‘‘active’’ SOE may be formed
using a 2D array of individually addressable light-emitting elements. In this case, each element in
the array is responsible for the production of voxels along a linear track. Alternatively, a ‘‘passive’’
SOE may be created by simply forming the screen from a translucent material able to scatter
incident light. Voxels may then be activated using one or more directed laser beams or through
the use of a DLP® Projector.
In the case of the former approach, voxel activation is essentially sequential, whereas in contrast
the DLP projector paradigm supports a high degree of parallelism (P). The voxel activation
capacity (NA) is given by:

Where:

• T denotes the average time required to activate each voxel and

• f the image refresh frequency
Additionally, it is useful to define a fill factor (ψ) such that:

Typically, to support the formation of image components at high resolution, the image space is
represented by a dense lattice of NL possible voxel locations, and parallelism in the voxel
activation system is the primary factor that determines the fill factor. Thus, in the case that voxel
activation is achieved using a DLP projector, all possible voxel locations can usually be addressed
during each refresh period. Conversely, voxel activation by means of a single directed laser beam
is a sequential process (P = 1) and typically results in a fill factor that is significantly less than 1%.
Electron beams can be deflected more rapidly than laser beams and the deflection apparatus is
both compact and cost effective. Consequently, the use of several directed electron beams
coupled with an SOE comprising one or more phosphor coatings, continues to provide a possible
low-cost solution to the implementation of a swept-volume display. Assuming that a single beam
is used, a high-performance electron gun equipped with electrostatic beam deflection apparatus
can achieve a voxel activation time on the order of 0.05 µs.
ii) Rotational Motion of a Planar Surface
In the case of this most widely researched form of volumetric display, the image space is formed
through the rotational motion of a planar screen (usually the axis of rotation is in the plane of its
surface). If an active SOE is employed then each element is responsible for the formation of voxels
along a circular track (the voxel bandwidth increasing linearly with radial distance from the axis
of rotation). Various approaches may be adopted in the transfer of control data to the rotating
SOE. These include the use of serial and parallel optical links, or possibly a wireless link.
Static voltages are usually conveyed using a traditional commutator. Alternatively, a ‘‘passive’’
SOE may be employed whereby one or both surfaces of the screen are able to scatter incident
light (voxels being created using directed laser beams or by projecting sequences of image slices
using DLP projector technology). A key implementation issue concerns the continually varying
geometry between the rotating SOE and the static voxel activation subsystems.
In the case of the Perspecta™ display >Fig. 6, an optical arrangement is used to maintain constant
geometry between the screen and DLP projector system.
An alternative approach is to use a screen coated with one or more phosphors – in which case
voxel activation is achieved using electron beams. An early radical embodiment maintained
constant geometry by corotating two electron guns with the screen. In contrast, Cathode Ray
Sphere prototypes employ either two or three static electron guns arranged around the periphery
of the image space – each gun being responsible for voxel activation across only a limited portion
of the screen’s motion (i.e., across the range of positions in which the geometry between SOE
and electron beam source is favorable). This ameliorates voxel elongation and the loss in voxel
positioning accuracy which occur when the surface of the screen lies at more acute angles to a
beam source.
For the most part, the rotational motion of a planar screen gives rise to an optically neutral image
space. However, when the plane of the screen lies at an acute angle to the user, a ‘‘visual dead
zone’’ may be observed. (In the context of volumetric systems, a ‘‘dead zone’’ refers to a region
of image space within which image quality is compromised by a reduction in one or more image
space characteristics.) The visual dead zone takes the form of a region of diminished image
intensity, and although its impact is ameliorated by binocular vision it can still cause a significant
distraction. The extent of this dead zone can be reduced by ensuring that the SOE and any
associated supporting structure are formed entirely from transparent components, and that their
thickness is minimized.
Prior to depiction, the volumetric data set is ordered into a series of radial slices (sectors), and
these are output in synchronism with the rotation of the SOE. In principle, during a single rotation
of the screen each voxel can be addressed twice, thereby enabling a doubling of the image refresh
frequency (this contrasts with the translational motion scenario in which the refresh frequency
must equal the frequency of motion of the SOE). However, in some embodiments, the
opportunity to activate voxels twice per screen rotation is exploited for multicolor image
formation.

Fig. 6: A CT scan of a human skull depicted on a Perspecta™ display. For applications such as this,
the translucent nature of the image is most desirable.

iii) Rotational Motion of a Helical Surface

When translational motion is used, mechanical considerations place an upper limit on image
space dimensions. In contrast, the rotational motion scenario enables a very simple drive
mechanism to be employed. However, the need to minimize the screen’s thickness while at the
same time not compromising its rigidity, imposes limitations on the image space diameter
(naturally effecting rotation within an evacuated vessel ameliorates this problem).
Alternatively, a helical screen configuration may be adopted and this provides a mechanically rigid
assembly able to support the formation of a large volume image space. Typically the helix is
bonded to a closed transparent cylinder – thereby increasing structural strength and preventing
undesirable air turbulence.
The use of a helical screen was pioneered by Ru˝digerHartwig in the 1970s, and since that time
the approach has received considerable attention. Embodiments to date have usually employed
a translucent light-scattering surface in conjunction with a plurality of laser beams for voxel
activation. The gradient of the helical surface gradually increases toward the axis of rotation and
this can result in unacceptable voxel elongation. This leads to a central dead zone, the extent of
which can be reduced by lessening the height of the helix and at the same time increasing its
diameter. Thus the helical configuration is most suited to the formation of a cylindrical image
space with a small height-to-width ratio.

b) Static-Volume Techniques
The static-volume approach places no reliance on mechanical motion for image space formation
and, in the most simplistic scenario, an image space could be constructed using a 3D array of
individually addressable light-emitting elements. Clearly, the number of elements needed for the
formation of an image space of useful dimensions and for the depiction of images of satisfactory
resolution would be excessive (a cubic image space with sides of length 40 cm and with an inter-
voxel spacing of 0.2mm would require 8×109 elements). Furthermore, doubling the linear
dimensions of the image space would result in an eightfold increase in voxel elements.
While this approach could, in principle, support the formation of a uniform and predictable
tableau with respect to image formation, the presence of large numbers of elements and
associated connections would result in the anisotropic propagation of light and so considerable
degradation in image quality. In contrast, in the case of a swept-volume system employing an
active SOE (moving with, for example, translational motion), the same inter-voxel resolution could
be achieved with only 4 × 106 elements.
Considerable research has been directed to the formation of an image space comprising a solid
or gaseous medium within which voxels are activated at the point of intersection of two
nonvisible, directed beams through the two-step excitation of fluorescence. Following the up
conversion process, return to the ground state (usually via intermediate energy states) leads to
the emission of visible radiation.
One key issue concerning the implementation of any display that utilizes the two-step excitation
process relates to the finite lifetime of the intermediate state. As the beam responsible for
stimulating excitation from the ground state is directed to different parts of the image space,
molecules on which it impinges will be excited to the intermediate state. If the second beam
happens to pass through a region(s) in which molecules have been inadvertently stimulated,
‘‘ghost’’ voxels will be formed.
This problem can be ameliorated by minimizing the lifetime of the intermediate state. The Depth
Cube technology employs an image space formed from a stack of twenty liquid crystal panels,
each of which can be individually switched between transparent and scattering states. Each panel
is separated from its neighbors by an appropriate amount, and a DLP projector is used to cast a
repeated sequence of image slices into the image space. As the projector sequences through the
image slices, the LCD panels are each briefly, sequentially switched to the scattering state. Thus
an image slice becomes visible at a depth within the image space as defined by the location of
the LCD panel that is in a scattering state at the time of its projection. This approach represents a
‘‘solid-state’’ version of a swept-volume display employing the translational motion of a planar
screen.
A further related technology employs a stack of transmissive LCD panels that are illuminated with
polarized light – in this embodiment, individual elements within each panel can be addressed.
A technique employing a 3D array of passive optical scattering centers formed in glass is indicative
of the considerable diversity of approaches that have been, and continue to be, the subject of
research. The array of scattering centers is created using a focused laser (laser induced damage
by local heating) and they are positioned so that each has unique coordinates in the horizontal
plane. Voxel activation is achieved using a collimated set of controlled laser beams which enter
the glass block from below. Thus each beam is able to activate a voxel at a certain location within
the image space.
c) The Ethereal Image Space
The exemplar display technologies outlined above support a physical image space, but the
techniques used in its formation preclude the insertion of physical interaction tools. Thus it is not
possible for the image and interaction spaces to coincide and so, for example, a haptic probe
cannot make direct ‘‘contact’’ with components that comprise the image scene. However, as
indicated in >Fig. 3, a ‘‘reimaging/projection’’ system can be used to cast the volumetric image
into free space thereby forming an ethereal image space, which in principle can be coincident
with the interaction space.
Optical projection systems of this type are by no means new and date back at least as far as the
early nineteenth century when, for example, the Megascope was used for 3D image projection as
part of the Phantasmagoria. The association of volumetric systems with an ethereal image space
offers to significantly advance the interaction process, but can only be satisfactorily achieved with
certain forms of physical image space. For an example of basic work undertaken in the projection
of volumetric images (created by the translational motion of an active SOE).
One radical approach to the direct formation of an ethereal image space employs a cloud of small
(light scattering) dust particles whose positions are continually monitored. When a particle passes
through a point in space corresponding to the coordinates of a voxel in the image data set, it is
briefly illuminated (naturally this assumes that the particle is in line of sight of the voxel activation
mechanism). The time at which each voxel is refreshed cannot therefore be defined in advance.
Furthermore, increasing the resolution of the display is likely to impact on the voxel refresh
frequency.
 d) Image Opacity
As indicated above, the great majority of volumetric display technologies researched to date have
given rise to translucent images. Although for some applications this may be advantageous, there
are many situations in which it is desirable to view, and interact with, opaque ‘‘solid’’ images. One
approach to implementing a display able to support the depiction of such images is based on the
formation of voxels which scatter ambient light. In this scenario, an image is only visible when it
is externally illuminated, and furthermore the position of external light source(s) determines the
form and location of natural shadows and shading.
In principle, this may be achieved through the use of photochromic materials which in essence
can be switched between two optical states by the application of suitable radiation. Although
such a material will then revert to its original state over time, this process can be expedited by
further application of radiation.
In summary:

Thus the application of radiation of wavelength λ1 causes a change from state A to state B.
Subsequent application of radiation of wavelength λ2 is intended to reverse this process. For
details of an investigation into the use of photochromic materials in connection with volumetric
systems.
The swept-volume display paradigm using the rotational motion of a planar screen can be
adapted for the generation of opaque images. In the case of one approach, the planar screen is
tilted at 450 to the axis of rotation and image sequences are projected onto it from above using a
DLP projector. The screen reflects the images in a horizontal direction (although the tilted screen
approach provides a simple means of projecting images, it unfortunately does not sweep out a
closed cylinder).
In order to create opaque images, the SOE is equipped with an optical diffuser. This has
anisotropic characteristics, being diffuse in the vertical direction and sharply specular in the
horizontal direction. Thus, in essence if a vertical strip was illuminated on the SOE and the screen
was slowly rotated, it would radiate a vertical column of light moving around the display volume.
To operate as a volumetric display, the screen must naturally rotate with sufficient rapidity
(thereby avoiding image flicker) and the DLP projection system must allow large numbers of views
to be cast onto it during the course of each revolution.
Observers located at different vantage points around the image space are effectively presented
with their own unique view onto the image scene.
As with flat-screen multi-view displays, this approach radiates a plurality of unique views onto an
image scene and moving horizontally provides the observer with a different vantage point. Thus
motion parallax is supported – in the horizontal direction. In contrast, there is no direct support
for motion parallax in the vertical direction, although this can be achieved by tracking an
observer’s position and adjusting the image content accordingly.
This general form of display also possesses key characteristics associated with the volumetric
approach – especially in respect of viewing freedom and the possibility of avoiding
accommodation/ convergence breakdown. Thus this approach can be regarded as the
amalgamation of volumetric and multi-view techniques and is best represented as constituting a
hybrid technology.

Summary
The conventional (monocular) display provides support for 3D image depiction by making use of
the pictorial set of depth cues. Consequently, images presented on such displays share many of
the visual characteristics that we associate with traditional forms of artistic painting. In contrast,
the spatial and visual characteristics of volumetric images have much in common with those
embraced by traditional sculpture. Similarities become increasingly evident when, for example,
we include opportunities for the formation of volumetric images within ethereal space, or for the
production of ‘‘solid’’ images made visible by the scattering of ambient illumination (thereby
enabling the formation of natural shadows and shading).
Advancing the traditional volumetric paradigm to support these, and various other, exciting image
depiction opportunities is readily within our grasp. However, vital work still remains to be carried
out in developing, optimizing, and integrating several key display and interaction subsystems –
thereby ensuring that they individually, and as a whole, operate synergistically and harmoniously
with the complex human sensory systems.

PSEUDO-VOLUMETRIC 3D DISPLAY SOLUTIONS

Introduction
Futuristic visions of mid-air images are frequently seen in TV commercials and sci-fi movies (see
>Fig. 1). Illusions of images floating in free space or 3D images that have real depth can be
generated in a variety of ways, but none of these are truly free-floating images. Volumetric display
create real 3D images instead of a 3D illusion, as the light is emitted from the actual 3D positions,
but the images are in a confined display volume and interaction with them is limited. There are
several ‘‘immaterial,’’ nonsolid mid-air displays such as water and smoke screens.
Fig. 1: A visionary mid-air display in the movie Star Wars: Episode IV A New Hope
A) Mid-air Display Technologies
i) The FogScreen™ Technology
The patented FogScreen™ mid-air projection screen is an emerging display technology that
enables high-quality projected images hovering in thin air, which the viewer can reach or walk-
through. It is basically a 2D screen, but it can be extended to a pseudo-3D or pseudo-volumetric
display.
The image quality on the FogScreen is optimal for a particle screen. A surrounding nonturbulent
airflow protects the injected thin particle flow (dry fog) from turbulence. As the inner fog flow
forms a thin fog plane, it enables high-quality image projections and a dry walkthrough
experience.
The FogScreen is a living surface. The fog flow tends to get more turbulent further away from the
device and starts increasingly to break up at the proximity of the floor. The human eye and brain
tend to correct the slightly shivering image as the viewer typically concentrates on the content,
not on the display medium.
The resolution degrades significantly if the viewing or projection angles are very oblique, as the
FogScreen image plane has a thickness of about 1 cm and thus the neighboring pixels blend with
each other. The screen resolution is thus spatially and temporally variable. It is the highest on the
upper part of the screen and when viewed toward the projector. The usable screen height is
typically 1–2 m. The image quality is adequate for most entertainment applications, and in the
future also for detailed information display applications.
The standard devices are suspended from ceiling or a truss (>Fig. 2), and the downward fog flow
creates 1 or 2 m wide screens (50” and 100” diagonal, respectively). Several units can also be
linked seamlessly together. Also, e.g., sideways flows or smaller reach-through desktop models
with upward flows are possible. The screens can be used outdoors only under calm wind
conditions.

Fig. 2: The FogScreen mid-air display is an intriguing experience

Tap water is broken into tiny fog particles so that they feel very dry and soon evaporate to air. No
chemicals are used. The vast majority of light is dispersed in the forward direction through the
micron-scale fog droplets (see >Fig. 3) and the colors are well preserved. Back projection is thus
recommended, as it produces much higher brightness than front-projection.
The light scattering properties of fog also enable one to project divergent images on both sides
without them interfering with each other.
Fig. 3: Angular intensity distribution of Mie scattered light on the FogScreen
By adjusting the fog density and image brightness, the screen can be made almost opaque, or
nearly fully transparent, in which case only the bright objects are visible and appear as if they are
floating in mid-air, as seen in >Fig. 4. This can be improved even further if the projector’s contrast
is high (especially laser projectors) and ambient light is low. The FogScreen can be used in brightly
lit environments like trade shows and exhibitions. A dark backdrop is recommended for better
effect.
The FogScreen can be made interactive with sensors, which react to proximity or ‘‘touch’’ of the
screen plane. Handheld pointer interaction is suitable for one master user, whereas plain hand
pointing is more natural and unobtrusive. However, plain hand pointing prevents the use of any
mouse buttons or similar tools, thus requiring more advanced interaction techniques.
The FogScreen is widely used in entertainment, theme parks, trade shows, special events,
museums, world expos, and nightclubs. It has also been in live TV shows, Hollywood movie
promotions, etc. It is currently an eye-catcher and a special effect, but it could also be used in a
much broader market as an immaterial information display, gaming or advertising platform, or as
a mid-air control panel for various devices such as washing machine.
The walk-through FogScreen can also employ virtual reality, mixed reality, and immersive
projection technology. Many kinds of intriguing pseudo-3D displays can be created and all
stereoscopic and virtual reality (>Fig. 4) techniques work with the FogScreen.
Fig. 4: A 3D model of hand bones on a virtual reality FogScreen
The stereoscopic effect is pronounced compared with traditional projection screens. Virtual
reality enables the viewer to freely choose his or her vantage point, but the image resolution and
intensity on the FogScreen quickly decrease as the viewer moves to the sides of the 2D screen, as
the pixels start to blend with their neighbors. These problems can however be reduced.
In order to create a volumetric display, using multiple parallel fog projection layers is not a good
approach, as each separate projection layer would need sufficient throw distance and projection
angle because the projected images can shine through to the next layer. However,
autostereoscopic viewing is possible to a limited extent. Multiple screen planes with depth fused
3D rendering can improve depth perception without any glasses, producing a pseudo-volumetric
mid-air screen.
The major advantages of the FogScreen are its walk-through and immaterial nature, translucency,
superior image quality, and larger screen compared with earlier particle screens, possibility for
direct interaction, and its very intriguing appearance. It encourages the audience to play with it
for extended periods of time. It is a magical experience if well done.
The permeability of the FogScreen enables new imaging, visualization, and user interface
possibilities, as the users can directly interact with the displayed mid-air objects and select and
manipulate them in a natural and intuitive manner without physical limits imposed by screens or
confined display volumes. The user occupies the same space as the image.
The FogScreen can even be described as a novel media platform as it has some fundamental
differences, possibilities, and advantages over other displays.
Other Non-solid Displays
There are numerous projection screens using water, smoke, fog, or cryofog (liquid nitrogen), but
their use is often very limited. The earliest example is the Ornamental Fountain dating back to
the end of the nineteenth century. More recently, water screens such as Jeep Waterfall,
Aquatique Show, and Disney’s Fantasmic spray sheets of freely flowing or pressurized water from
nozzles, or make huge amounts of high-velocity water collide to a metal plate in order to create
impressive foggy screens for large audience. The magnitude and wetness of these screens make
them impractical for indoor, walk-through, or small-scale applications. However, many of these
water screens may look fantastic if viewed from afar and on-axis at night.
Many primitive fog screen systems have been used as a special effect in art and entertainment,
but the fidelity of the projected images is low, as they do not use the laminar FogScreen
technology. The fog disperses rapidly due to friction and dynamic pressure difference in the flow,
which disrupts the desired smooth planar surface and severely distorts the image. The image
quality on FogScreen is significantly better than with any previous methods. >Figure 5 shows
screens consisting of fog with and without the protective laminar airflow.
Very charming experimental systems like soap bubbles, volumetric water display, and water
domes have been presented. The imagination of artists and engineers seem to be limitless.

Fig. 5: The projection resolution with and without the protective laminar flow. A traditional fog
screen flow is inferior to the FogScreen™. The viewing angle is at 450 on the side.

Exotic Volumetric and Other Displays

Transparent projection screens are acrylic or glass plates which are coated with a holographic
film. The viewer sees only the objects and light which are projected in a 35 0 angle, and the rest
of the screen is transparent. The result is a bright and clear image apparently floating in air,
although no walk-through is possible. It is used as an eye-catcher in window exhibitions, etc.
Impressive ‘‘hologram’’ setups.
Cheoptics 360 uses transparent screens in a pyramid-shaped construction. It can show images
apparently floating in thin air, but again, the screens are not penetrable, and the whole
construction takes a lot of space.
Laser shows with haze in air are used as attractions at theme parks and concerts. The laser beam
itself is not visible, but the light reflecting from dust or particles in the air can be seen.
Another laser technique creates an illusion when laser beam hits black scrims in a dark
environment, so that the scanned laser image appears to float in mid-air, since only the laser-
illuminated part of the scrims can be seen.
Another technique uses laser beams to project a volumetric image on to a display medium which
may consist of a volume of gas, liquid, aerogel, or a solid glass cube. The image appears to float
inside the volume without the need of stereoscopic glasses, and the eye can properly
accommodate to it. Tiny glass image cubes based on gated, two-frequency, up-conversion
fluorescence can be viewed from almost any angle by multiple viewers. Their writing speed is low,
the cubes are extremely dim even in a darkened room, not penetrable, and very small (or
alternatively large glass cubes would weigh a lot).
A volumetric plasma emission display employs short, strong laser pulses which are focused to a
point in air. The air turns into hot glowing plasma and produces a very loud and bright flashpoint
for a fraction of a second. The method is slow, no colors are possible, it produces ozone and it can
be dangerous (no touching!). The device produces only 120 dots per second, which at 25 Hz
refresh rate gives 5 voxels per frame. To increase the frequency, also higher output energy is
needed. It has big inherent problems and is hardly practical in the presented form.
Phantasmagoria and its projected ghost illusions created some hysteria in the late eighteenth
century. Also Pepper’s Ghost is an old illusion sometimes used in theaters. It uses a
semitransparent glass and varying illumination between the direct view and the reflected view.
Also various configurations of concave spherical or half mirrors, or head-up displays can be used.
These displays can create an illusion of mid-air images, but they have constraints on viewing
angles, depending on the configuration.
Holodust is another idea, which scans and illuminates flowing dust particles in midair by laser. A
patent is granted for it, but it seems to be only an unimplemented idea. No proof of- concept
prototypes exist to this date. Holodust may be very complex to be built in full color, and it requires
very high-speed scanning and projection systems.
TerrainTable is a ‘‘2½D volumetric’’ visualization table that enables users to see and interact with
a 3D terrain model. Using an array of vertical pins beneath a silicone skin, the table can create
curved surfaces within a limited tabletop area. It employs an overhead projector, and makes
convincing topographical maps.
Also 3D printers which create tangible 3D objects by spraying numerous thin plastic layers could
be seen as one subcategory of 3D displays.

Conclusions
We have briefly described some technologies to create mid-air or niche volumetric displays. Any
display technology has its strengths and weaknesses. Most 3D displays are not mid-air displays,
even though they may seem like that from afar. Interaction with the current volumetric displays
is limited, as they are not reach-through and no direct contact with the objects is possible.
The human vision is designed as a highly delicate, complex, and sophisticated sensory fusion
system. Even so, our visual system has some perceptual shortcomings which can be taken
advantage of by using tricks and approximations.
A perfect ‘‘floating 3D holographic free-space projection’’ may never be possible, but neither is it
required. For any 3D or midair display, a demand must meet a suitable price. For any technology
to penetrate the markets, also many nontechnical issues such as revenue, marketing, consumer
needs and acceptance, intellectual property rights, backward compatibility, price, timing, and luck
are paramount.
In many respects, the FogScreen seems to be a feasible shortcut technology to create a
‘‘holographic,’’ Star Wars-like mid-air display, as it is truly reach-through, not restricting user’s
movements and interaction. In the future, improved laser, LED and pocket projectors, mid-air VR
and 3D experiences, and novel mid-air user interfaces may enable many daily life applications for
the mid-air screens. Mid-air displays have initially been a novelty, but in the long term they could
have a big impact also on application areas such as CAD, data visualization, digital signage, tele-
presence, simulation, and entertainment.