C H A P T E R 1

Depth Cues

Introduction

This handbook is about getting the most out of StereoGraphics products by acquiring a background in the fundamentals: how people see three-dimensionally and stereoscopically (they're not necessarily the same thing, as we shall learn), how electro-stereoscopic displays work, and how they interface with the visual system. While guidelines for creating good-looking stereoscopic images are given, and aspects of the subject may be codified, the creation of a stereoscopic image, like the creation of any image, is an art.

The purpose of this guide is to enhance the appearance and usefulness of stereoscopic images which may be displayed by a monitor or a projector. The images may be created with a graphics workstation or personal computer.

A stereoscopic image presents the left and right eyes of the viewer with different perspective viewpoints, just as the viewer sees the visual world. From these two slightly different views, the eye-brain synthesizes an image of the world with stereoscopic depth. You see a single not double image, since the two are fused by the mind into one which is greater than the sum of the two.

A familiar example of a stereoscopic display is the View-Master ® viewer illustrated here. If you close one eye when looking into the View-Master stereoscope, the image looks flat. With both eyes open, you see an image with stereoscopic depth. This kind of image used by the View-Master or by our CrystalEyes or SimulEyes VR products is called a plano-stereoscopic image, because it uses two planar (flat or two-dimensional) images.

A 3-D database may be displayed stereoscopically and many graphics packages or games may be upgraded to produce a true binocular stereoscopic image. In most cases, this is best left to the software developer who has access to the source code. Moreover, the developer is best able to support his or her product. Experience has shown that adding stereoscopic capability to an existing package is straightforward - a word not used lightly in a world where software development often requires heroic efforts. All computer-generated images produced from a three-dimensional database require the computation of an image from a single perspective point of view or location in space. A stereoscopic image differs only in that the image must be generated from two locations.

As an alternative, the image may be captured from the real world by means of a camera. In other words, it may be photographed or videographed. Whether the image is computed or photographed, the generation or capture of images from two perspectives - a left and a right - is required. If a single perspective viewpoint can be produced, it is conceptually straightforward to create the additional perspective. The creation of such images is the subject matter of this handbook.

People in the field of interactive computer graphics have also been playing with language by using the terms "3-D" or three-dimensional to mean a realistic-looking image which may be appreciated with one eye. Unfortunately, this is confusing nomenclature because most people who speak English outside of the field of computer graphics use the term "3-D", and on occasion three-dimensional, when they mean a stereoscopic image. Such an image requires the beholder to use both eyes. The usual so-called 3-D computer image loses nothing when you close one eye, but a true stereoscopic image loses its raison d'etre when viewed with one rather than two eyes.

Monocular Cues

The monocular, or extrastereoscopic, depth cues are the basis for the perception of depth in visual displays, and are just as important as stereopsis for creating images which are perceived as truly three-dimensional. These cues include light and shade, relative size, interposition, textural gradient, aerial perspective, motion parallax and, most importantly, perspective. A more complete description of depth perception may be found in a basic text on perceptual psychology.1

Images which are rich in the monocular depth cues will be even easier to visualize when the binocular stereoscopic cue is added.

Light and shade provide a basic depth cue. Artists learn how to make objects look solid or rounded by shading them. Cast shadows can make an object appear to be resting on a surface.

Bright objects appear to be nearer than dim ones, and objects with bright colors look like they're closer than dark ones.

Relative size involves the size of the image of an object projected by the lens of the eye onto the retina. We know that objects appear larger when they are closer, and smaller when they are farther away. Memory helps us to make a judgment about the distance of familiar objects. A person seen at some great distance is interpreted to be far away rather than small.

Interposition is so obvious it's taken for granted. You perceive that the handbook you are now looking at is closer to you or in front of whatever is behind it, say your desk, because you can't see through the book. It is interposed between you and objects which are farther away.

Textural gradient is the only monocular depth cue articulated by a psychologist in modern times. The other cues were known and used by painters by the time of the Renaissance. A textured material, like a grassy lawn or the tweed of a jacket, provides a depth cue because the texture is more apparent as the object is closer to the observer.

Aerial perspective is the diminution in visibility of distant objects caused by intervening haze. Often, distant vistas will pick up a bluish haze because of the scattering of red light by the intervening atmosphere. In thick fog or haze, objects may not be all that distant.

Motion parallax is a depth cue which can be provided in the visual world or in moving-image displays, and can be achieved, for example, through the rotation of an object. A familiar example of this cue is seen from a moving car. Everyone has noticed that telephone poles move past more rapidly than the distant hills.

Not all graphic compositions can include motion parallax, but the use of motion parallax can produce a good depth effect, no matter what the spatial orientation of the axis of rotation. This axial independence indicates that motion parallax is a separate cue from stereopsis. If the two were related, only rotation about a vertical axis would provide a depth effect, but even rotation around a horizontal axis works. The point is made because some people think stereopsis and motion parallax are manifestations of the same entity. Motion parallax - as in the rotation of an object, for example - is a good cue to use in conjunction with stereopsis because it, like other monocular cues, stresses the stereoscopic cue. However, details or features of rotating objects are difficult to study.

Perspective, sometimes called geometric, rectilinear, or photographic perspective, is the most important extrastereoscopic depth cue for our purposes, since it scales the stereoscopic depth cue..2 Computer-generated images with a strong perspective cue look deeper and may be easier to view, as we shall learn in later chapters. Perspective is the relationship between foreground and background objects. If it is exaggerated, or if there are perspective cues such as lines receding to a vanishing point, the image's depth will be enhanced.

Depth Cuing is the name of a technique used in computer graphics. Depth cuing reduces the intensity of the object in proportion to the distance from the viewer.

Summing Up

Monocular depth cues are part of electronic displays, just as they are part of the visual world. While the usual electronic display doesn't supply the stereoscopic depth cue, it can do a good job of producing a seemingly three-dimensional view of the world with monocular cues. These depth cues, especially perspective and motion parallax, can help to heighten the stereoscopic depth cue.

References

1. Kaufman, Lloyd. Sight and mind: An introduction to visual perception. New York: Oxford University Press, 1974.

2. MacAdam, D.L. Stereoscopic perceptions of size, shape, distance and direction. SMPTE Journal, 1954, 62:271-93.

Chapter 2

Chapter 3

Chapter 4

Chapter 5

Glossary