Binocular Imbalance and Depth Perception in Visual Display Systems: a Set of Notes

1. Notes adapted from a report by B. A. J. Clark, 1996.

In 1970, a RAAF UH-1 Iroquois was on the ground with rotors turning when another Iroqouis taxied alongside it to refuel. The three aircrew in the second helicopter visually confirmed main rotors separation of about 1 metre. As the second Iroquois settled, the rotor clearance proved to be negative by about 1 metre instead and the rotors intermeshed destructively.

Also in 1970, the Standards Association of Australia began work on a world-first national standard for sunglasses. One of the safety issues driving the issue of a standard was that inequality in the light transmittance of the lenses of a pair of sunglasses could produce an illusory displacement of any object moving across the field of view (Clark 1968). The original form of this illusion was discovered by Pulfrich in 1922 and is a consequence of a relative visual latency differential in seeing by the eye receiving the reduced amount of light.

Illusory alteration of apparent distance and angular speed of objects viewed through the side window of a moving vehicle can be obtained by observing with a filter in front of one or other eye. This effect can be seen from aircraft, road vehicles and boats, especially at night through night vision goggles (NVGs) with one or other side filtered to produce a gain imbalance exceeding the 10% (approx.) tolerance adopted in the major national standards for sunglasses. Gain imbalance has been mentioned several times by aviators as a factor in NVGs flight incidents (Crowley, Rash and Stephens 1992).

If an aircrew person's right eye is slightly faster than the left eye in seeing (an anisopia effect), an ownship Iroquois rotor seen from near its axis with the blades moving from right to left could be expected to appear smaller in radius than a stationary blade. The rotor disk on the adjacent Iroquois would likewise appear to be foreshortened by more than the amount caused by the oblique view. If the Pulfrich illusion does apply with fast moving blades, a 1-m perceptual error in the clearance between rotors would be caused by an interocular differential of only a few milliseconds.

An experimental apparatus was built to test this possibility. It had a two-bladed rotor, about half-scale for an Iroquois. A variable speed electric motor drive was provided, with maximum rotor rotational speed being about half that of a typical helicopter. The rotor disk appeared above the observer, as with a real helicopter. The disk was seen against an illuminated featureless background. The Pulfrich effect was observed: reversing either the filter position from one eye to the other or reversing the direction of blade rotation made the rotor disk look either larger or smaller. However, these effects took place simultaneously with a previously unknown illusion in which the rotor consistently looked either smaller or larger to individuals, the effect increasing with rotor speed but apparently independent of rotor direction. This new illusion was also visible through sunglasses and NVGs.

These illusory effects are in conflict with the monocular cues to distance. At night, the monocular cues are weakened and the illusions can become stronger.

The Pulfrich Effect is a binocular illusion. It takes the form of a distortion of the apparent path of a pendulum swinging in a fronto-parallel plane when the observer has a natural (anisopia) or induced imbalance between the illuminance and/or sensitivities of the retinas of the two eyes, as with a neutral density filter in front of one eye. The pendulum appears to follow a path that is near elliptical in plan. The direction of motion, clockwise or anticlockwise, depends on which eye has the slower response. The effect is generally accepted to be a result of a differential delay in transmission time of information from retina to brain, arising because the signal transmission time increases monotonically with reducing retinal illuminance. In the case of objects with a component of apparent motion across the retinas, the brain interprets the between-eyes image differences as stereoscopic disparities, leading to perceptual conflicts between the monocular and binocular cues to distance.

The Pulfrich Effect is readily demonstrable through binocular optical instruments, including NVGs, by using a neutral density filter either in front of one objective or between one eye and the eye lens. The existence of perceptual conflicts between monocular, biocular and binocular views through NVGs with poor matching of light gains between barrels may justify reappraisal of helo-NVGs accident cases of collision with obstacles (also in Crowley, Rash and Stephens 1992).

The 'linear Pulfrich Effect' is observed by looking sideways, with a filter covering one eye, out of a moving vehicle. With the filter over the leading eye, the ground plane appears lower, objects more distant along the line of sight and apparently faster than actual. Reversing the filter position or motion raises the apparent ground plane along the line of sight and makes objects appear closer and slower.

The rationale for sunglasses lens matching tolerance: (0.06 in density, Clark 1968) is considered applicable as a first step in controlling the problem, so that the output luminance match between the two channels of any biocular or binocular display system should be +-13% as a go/nogo limit, pending fuller investigation.

Clark, B. A. J. (1968) The luminous transmission factor of sunglasses. American Journal of Optometry and Archives of the American Academy of Optometry, 46, 362- 378.

Crowley, J. S., Rash, C. E. and Stephens, R. L. (1992) Visual illusions and other effects with night vision devices. Pp 166-180, Helmet Mounted Displays III, Ed. T. M. Lippert, Proceedings of the Society for Photo-Optical Instrumentation Engineers No 1695, Washington, USA: SPIE.

Enright, J. T. (1970) Distortions of apparent velocity: a new optical illusion. Science, 168, 464-467.

2. Extract from Clark, B. A. J. (1996) Mismatches between driver visual capabilities and road vehicle standards. Road and Transport Research, 5 (2), 92-117.

This extract is about a formula derived from observations of perceptual delays in one subject using a neutral density filter in front of one eye. Data pooled from several subjects should be sought from the more recent literature if greater reliability in prediction of effects is desired. The example given shows how the formula can be applied to calculate the difference in perceptual latency when observing with two eyes through cosmetic tinted lenses or similar when the two lenses have differing transmittances. The horizon sky passes through a luminance of 1.0 cd/m² during twilight.

In the so-called Hess effect¹ and the Pulfrich (1922) effect², the perceptual latency of seeing t increases monotonically as the retinal illuminance decreases (Guth, 1964): no cone-rod discontinuity is apparent. For a subject of Rogers and Anstis (1972), with luminance L instead of troland value (Wyszecki and Stiles, 1967),

The perceptual latency at 1.0 cd/m², as viewed through a filter with T = 88% is 46.0 ms and, with T = 66%, 48.5 ms. The latency differential is thus 2.5 ms. At 0.01 cd/m², as in rural night driving, the differential is 4.1 ms.

ROGERS, B. J. and ANSTIS, S. N. (1972) Intensity versus adaptation and the Pulfrich stereo phenomenon. Vision Research, 12, pp 909-928.

PULFRICH, C. (1922) Die Stereoskopie im Dienste der isochromen und heterochromen Photometrie. Die Naturwissenschaften (Berlin), 10, pp 553- 564, 569-574, 596-601, 714-722, 735-743 and 751-761.

WYSZECKI, G. and STILES, W. S. (1967) Color Science. Wiley, New York, USA.

1. An illusory phase displacement is observed monocularly when images of different luminance's are oscillated together. The dimmer image lags because of perceptual latency. Williams and Lit (1983) called this the Hess effect even though they cited earlier studies by Szili and Charpentier.

2. Pulfrich effect: an illusory near-elliptical motion of a pendulum bob swinging in a fronto-parallel plane and observed binocularly with one eye covered by a filter. The longer latency in the visual pathway from the dimmer retinal image is largely responsible (Lit, 1949; Wilson and Anstis, 1969; Spafford and Cotnam, 1989; Emerson and Pesta, 1992). The effect can produce illusory localisations of traffic and roadside objects (eg Allen, 1961; Clark, 1969; Enright, 1970; Landrigan, 1984).

3. Extract from notes relating to evidence given in public by Dr Barry Clark to the enquiry into the mid-air collision of two Australian Army Blackhawk helicopters, 1996

Illusory alteration of apparent distance and angular speed of terrain objects viewed through the side window of a moving vehicle can be obtained by observing with a filter in front of one or other eye. This effect, analogous to the Enright effect, has been seen by numerous others in company with the writer on many occasions from aircraft, road vehicles and boats, for relative azimuths determined to be within the ranges of between about 1 to 5 o'clock and 7 to 11 o'clock. Objects such as other vehicles with their own motion appear to be displaced in depth regardless of relative azimuth, provided that that they have an appreciable component of relative motion across the field of regard, ie transverse motion.