To achieve stereopsis it is necessary to generate two images and present them separately to each eye. Depending on the purpose of the exercise, the images may or may not be correlated. In an ideal world too there should be no timing skew between the two images as the processing paths from the two eyes are more sensitive to timing disparities than would be suggested by considering the temporal bandwidth of the overall system. There are several ways to do this, each of which has its own disadvantages.

5.2.1    Two stimulus generators

The simplest way to create two stimuli is to have two distinct stimulus generators and display devices and use an optical system to ensure that the image from each impinges only on the intended eye. This works well if the subject can be fixed rigidly in a chair and a septum is used to prevent crosstalk. It may be difficult to arrange however if the display devices are large as it may not be possible to get them very close together. In addition, there remains the problem that the two displays have to be adjusted to have the same geometry, brightness, luminance and colour and some mechanism needs to be found to synchronise the two frame scans not to mention the two generators.

5.2.2    Split screen

A variation on the two stimulus generators method above is to use one stimulus generator with a large display and split the screen into two areas and again using a septum to prevent crosstalk. The advantage here is that there is no need to solve the synchronisation problems and that any adjustments for geometry and luminance will apply equally to both images. Even though this technique halves the available viewing area you can be quite sure that there is no timing difference between the two eyes.

5.2.3    Shutters

The third method of simulating stereopsis and that most often practised is to use a pair of shutter goggles synchronised with the display. The idea is to present the images destined for each eye on alternate frames of the display and then prevent the light reaching the wrong eye by closing a shutter in front of it. The ideal characteristics of these shutters should be instantaneous switching from open to closed and vice-versa, 100% distortion free transmission in the open state and 0% transmission in the closed state and they should be light enough to be worn either as goggles or made into spectacles. In the past, these have been constructed of black rotating discs that have apertures cut out of the sides driven synchronously with the display rather similar in concept to the shutter systems on movie projectors. While having very good optical characteristics their portability qualities leave room for improvement: not everyone likes the idea of something akin to a circular saw blade two inches from their face. In recent years the emergence of liquid crystal materials has provided an appropriate non-moving alternative. These rely on the property that some liquids have of being able to rotate the plane of polarisation of light when a small electric field is applied. When sandwiched between a pair of crossed polarisers and connected to suitable voltage source they can form the basis of a shutter system light enough to be worn on the head. Currently, the best available system uses a ferro-electric material that switches state in less than 100microseconds and can be driven from a simple 5V signal source. In the open state, transmission is 25% while when closed it may be as much as 3log units lower. The closed state transmission relies heavily on the accuracy with which the polaroid sandwich has been aligned and shows very substantial degradation at elevated temperatures and off the principle axis. When expressed as a contrast, an open to closed transmission ratio of 1000:1 sounds very impressive but will undoubtedly fail to impress a dark-adapted observer who expects to be able to see nothing through them when they are closed. The primary sources of error when using this stereoscopic system are a guaranteed eye to eye temporal disparity of one display frame time and crosstalk from one eye to the other caused by the finite extinction ratio of the goggles and the decay time of the display phosphor. It is possible to buy monochrome CRTs with a short persistence phosphor for this purpose but in practice, the deficiencies of the goggles are the dominant factor. Whether the crosstalk is invisible or important will depend on the exact stimuli being used. Remember that the eye is sensitive to local contrasts rather than luminances so that if one eye is presented with a high contrast pattern while the other eye is shown a black image crosstalk is almost inevitable. (To check the crosstalk in a system display a stimulus such as a plaid that is constructed by interleaving frames and close one eye and see how much of the image intended for the other eye is visible.) Don't forget also when using this system that the net frame rate of the image presented to each eye is half the total display frame rate. If the display is operating at 140Hz each eye will see an image presented at only 70Hz and consequently may seem somewhat flickery.

^ Back to top