Stereoscopic and 3-D images create or present spatial depth through coordinated image information. In stereoscopic systems, two slightly different images are presented to the left and right eyes so that binocular disparity can contribute to the perception of depth. Other 3-D image and display systems may use optical, physical, computational or display-based methods to produce spatial effects.
This visual collection presents representative stereoscopic and 3-D images, including stereo pairs, anaglyphs, polarized images, lenticular and autostereoscopic displays, stereograms, 3-D photography, Virtual Reality and computer-generated stereo imagery. The examples show that a 3-D image may reproduce, simulate or generate depth through several very different processes.
Stereoscopic and 3-D images
Explore representative examples below. Each image illustrates a particular form, application or visual principle of stereoscopic and three-dimensional imaging.
Stereoscopic Image Pairs
A stereoscopic pair contains separate left-eye and right-eye images representing slightly different viewpoints. When presented correctly to the corresponding eyes, the differences between them can generate binocular stereoscopic depth.
Anaglyph 3-D Images
Anaglyph images encode the left- and right-eye views using different colour channels. Coloured filters worn over the eyes separate the two image components so that each eye receives principally its intended view.
Polarized 3-D Images
Polarized stereoscopic systems superimpose two image views while using different polarization states to separate them. Corresponding filters in the viewer’s glasses direct the appropriate image to each eye.
Lenticular & Autostereoscopic Images
Lenticular and other autostereoscopic systems direct different image information towards different viewing positions. They can produce stereoscopic or multi-view depth effects without requiring conventional 3-D glasses.
Stereograms
Stereograms encode binocular depth relationships within an image or image pair. Some require separate left- and right-eye views, while autostereograms can conceal a stereoscopic depth structure within a single repeated visual pattern.
Stereoscopic Photography
Stereoscopic photography records a scene from two horizontally separated viewpoints corresponding approximately to the positions of the two eyes. The resulting image pair can then be viewed stereoscopically.
Virtual Reality Stereoscopic Views
Virtual Reality headsets present separately rendered views to the two eyes while updating them as the head moves. Stereoscopic disparity is therefore combined with changing viewpoint and immersive field of view.
Computer-Generated 3-D Images
Computer graphics can generate coordinated left- and right-eye images from virtual cameras, allowing stereoscopic depth to be produced for cinema, games, simulation, scientific visualisation and immersive environments.
What stereoscopic images show
Stereoscopic images provide different but coordinated image information to the two eyes. Because the left and right eyes occupy different positions, each receives a slightly different view of a nearby three-dimensional scene.
The positional differences between corresponding image features are known as binocular or stereoscopic disparities. When appropriate disparities are presented to the two eyes, the visual system can use them as information about relative depth.
Stereoscopic imaging therefore reproduces one important binocular component of natural spatial vision rather than merely drawing depth cues onto a single flat image.
Read the full Binocular Vision page →
The left-eye and right-eye images
A stereoscopic image normally begins with two views. One represents the scene from a left-eye position and the other from a right-eye position.
The two images are similar but not identical. Objects at different depths occupy slightly different relative positions in each view because each eye observes the scene from a different spatial location.
The method used to deliver these two views to the correct eyes may vary considerably. The pair may be viewed through a stereoscope, separated by colour, polarization or shutter timing, directed optically by a display surface, or presented on separate screens within a headset.
Binocular disparity and stereoscopic depth
Binocular disparity describes the difference between corresponding positions in the two retinal or presented images. The amount and direction of disparity vary with the spatial relationship between the observer and the viewed objects.
Stereoscopic systems deliberately reproduce or generate these disparities. The visual system then combines the two views into a single perceptual spatial experience in which some elements may appear in front of, on or behind the apparent image plane.
The strength and usefulness of stereoscopic disparity are greatest at relatively close viewing distances and decline as the angular difference between the two eyes’ views becomes smaller with increasing distance.
Stereoscopic images and ordinary perspective images
An ordinary photograph or graphical perspective image presents one image to both eyes. It can contain strong monocular depth information through diminution, foreshortening, convergence, occlusion, atmosphere, shading and many other relationships, but the image itself does not provide separate left- and right-eye views.
A stereoscopic image adds binocular disparity by providing two coordinated views. This does not replace the ordinary perspective structure within each image: each member of the stereo pair may itself contain linear, curvilinear, photographic or other perspective relationships.
Stereoscopic Perspective can therefore be understood as an additional binocular relationship combined with the perspective structure of the individual images.
Parallel and cross-eyed stereo viewing
Some stereoscopic pairs can be viewed without a dedicated stereoscope. In parallel viewing, the left image is presented to the left eye and the right image to the right eye while the viewing directions remain comparatively parallel.
In cross-eyed viewing, the image positions are exchanged and the viewer converges the eyes so that each eye attends to the opposite member of the displayed pair.
Both methods can generate stereoscopic depth when the image geometry and viewing arrangement are appropriate, although they place different demands on ocular convergence.
Anaglyph, polarized and shutter 3-D systems
Many stereoscopic display systems place the two eye views within the same physical display area and then use a separation mechanism to ensure that each eye receives its intended image.
Anaglyph systems separate the images by colour. Polarized systems separate them by polarization. Active-shutter systems alternate the left- and right-eye views through time while electronically controlled glasses alternately block and reveal the corresponding eye.
The separation method changes, but the underlying stereoscopic principle remains the same: two coordinated views must reach the two eyes separately.
Autostereoscopic and lenticular images
Autostereoscopic displays attempt to provide binocular or multi-view depth without requiring the viewer to wear glasses. Lenticular lenses, parallax barriers and related optical structures direct different image information towards different viewing positions.
A viewer positioned within an appropriate viewing zone can therefore receive different images in the left and right eyes. More complex systems can provide several views so that the displayed appearance changes as the observer moves laterally.
Such systems connect stereoscopic imaging with motion parallax and viewpoint-dependent 3-D display.
Stereograms and hidden-depth images
A stereogram encodes stereoscopic relationships so that depth appears when the two eyes establish the intended correspondence between image features.
Traditional stereograms use separate image pairs. Random-dot stereograms demonstrate that stereoscopic form can be perceived from binocular disparity even when recognizable object outlines are absent from either monocular image.
Autostereograms go further by embedding disparity relationships within a single repeated pattern. The apparently flat pattern can then reveal a hidden spatial form when viewed with the required binocular relationship.
Stereoscopic photography and cinema
Stereoscopic photography records separate left- and right-eye views using two cameras, a dual-lens camera or successive exposures from displaced camera positions. The separation between the camera viewpoints provides the photographic stereo base.
Stereoscopic cinema applies the same principle to moving images. Two coordinated image sequences are recorded or generated and subsequently delivered separately to the viewer’s two eyes.
The choice of camera separation, convergence, focal parameters, screen size and viewing distance can all affect the scale and comfort of the resulting stereoscopic space.
Explore Perspective in Photography →
Virtual Reality and binocular 3-D
Virtual Reality headsets provide separate images directly to the two eyes and can update these images in response to head position and orientation. The viewer therefore receives binocular disparity together with changing visual direction and viewpoint.
This combination distinguishes immersive stereoscopic viewing from a fixed stereo photograph or cinema screen. The surrounding visual field can respond dynamically as the user looks and moves around.
Virtual Reality may consequently combine stereoscopic imaging, wide-field or spherical perspective, motion parallax, computer-generated perspective and interactive viewing within one system.
Explore Computer Graphics, Games & Extended Reality →
Stereoscopic images and broader 3-D images
The expressions stereoscopic image and 3-D image should not always be treated as synonymous. Stereoscopy is one specific method for producing visual depth through separate binocular views.
The wider category of 3-D imaging can also include multi-view and autostereoscopic images, volumetric displays, holographic or hologram-like systems, physically modelled three-dimensional displays and computer-generated spatial environments.
Some systems therefore create a stereoscopic impression of depth on or around a nominally planar display, while others provide different image information as the observer moves or place luminous or optical information at different physical or apparent spatial positions.
Read the full 3-D Display Systems page →
3-D image, 3-D representation and 3-D display
The term 3-D image can describe several different things and should therefore be used carefully. A conventional perspective drawing may depict a three-dimensional object while remaining physically two-dimensional. A stereoscopic display provides different two-dimensional images to the two eyes and produces a binocular spatial effect. A volumetric or other genuinely spatial display may locate visible information within a physical three-dimensional region.
It is therefore useful to distinguish the three-dimensional subject represented, the dimensional organisation of the image information, the optical process through which depth is presented, and the physical dimensionality of the display itself.
These distinctions prevent very different technologies from being grouped together merely because all are described informally as “3-D”.
Stereoscopic depth and other depth information
Binocular disparity is only one source of visual information about three-dimensional space. A stereoscopic image can simultaneously contain diminution, foreshortening, occlusion, shading, texture gradients, atmospheric effects, movement and familiar perspective convergence.
These monocular and binocular relationships may reinforce one another, but they can also conflict. A stereoscopic object can, for example, be assigned binocular depth that differs from the spatial relationships implied by its size or perspective construction.
The resulting 3-D experience therefore depends upon the coordination of several perspective and perceptual relationships rather than upon stereoscopy alone.
What to look for in a stereoscopic or 3-D image
- separate left-eye and right-eye views;
- small positional differences between corresponding image features;
- binocular or stereoscopic disparity indicating relative depth;
- whether the stereo pair is arranged for parallel or cross-eyed viewing;
- colour separation in an anaglyph image;
- polarization or shutter separation in stereoscopic displays;
- lenticular, parallax-barrier or other autostereoscopic image separation;
- hidden binocular depth within a stereogram or autostereogram;
- paired photographic or computer-generated camera viewpoints;
- whether the image appears in front of, on or behind the apparent display plane;
- whether viewpoint changes alter the displayed image;
- whether the system provides two views or multiple views;
- whether the image is stereoscopic, autostereoscopic, volumetric, holographic or another form of 3-D display;
- the distinction between a two-dimensional image representing a 3-D object and an image system that actually provides binocular or spatial depth information; and
- how stereoscopic depth interacts with ordinary perspective, occlusion, scale, shading and movement.
A stereoscopic or 3-D image should therefore be identified through the method by which depth information is produced and presented rather than simply because the depicted subject appears three-dimensional.