User Experience with
Partial Binocular Overlap
In Augmented Mixed Reality

Partial Binocular Overlap in AR
● Reminder: Binocular vision and Partial Binocular Overlap
● What are the problems with PBO in AR?
○ Previous research and our findings so far
● How bad is it?
○ What is bad and how to measure it? Existing metrics and our ideas.
● What are the optimal parameters within possible limits?
● What can be done to alleviate the problem areas?

Partial Binocular Overlap in AR
● Reminder: Binocular vision and Partial Binocular Overlap
● What are the problems with PBO in AR?
○ Previous research and our findings so far
● How bad is it?
○ What is bad and how to measure it? Existing metrics and our ideas.
● What are the optimal parameters within possible limits?
● What can be done to alleviate the problem areas?

Field of View and binocular overlap
Why wide FOV
● Situational awareness
● Embodied cognition
Why binocular vision
● Fast and accurate scene layout
perception

Perception of Space and Motion
● Binocularity is a secondary cue.
● Sensor fusion: multiple coherent
cues help with depth perception.
from Cutting, Vishton 1995
(http://people.psych.cornell.edu/~jec7/pubs/78.pdf)

from Cutting, Vishton 1995, Perception of space and Motion, p102 (http://people.psych.cornell.edu/~jec7/pubs/78.pdf)

Binocular perception beyond the eye

How Immersive Is Enough? A Meta-Analysis of the
Effect of Immersive Technology on User Presence.
JAMES J. CUMMINGS and JEREMY N. BAILENSON 2015
“...results show that increased levels of user-tracking, the use of stereoscopic
visuals, and wider fields of view of visual displays are significantly more impactful
than improvements to most other immersive system features, including quality of
visual and auditory content.”

What is the minimum FOV required for Navigation?
(Hassan, 2007)
FOV size strongly affects navigation task.
Binocularity does not.

In this research, a Unity project consisted of a moving object with variable
parameters was created to examine if there’s correlation exists between
players’ head direction and gaze direction in eye’s smooth pursuit movement.
Furthermore, object parameters, shape, color, distance, speed and horizontal
moving degree were tested to explore whether they can elicit statistically
significant differences in gaze prediction. Results revealed that while smoothly
pursuing a moving object with the gaze, people’s horizontal and vertical
component of head direction and gaze direction are separately linearly
correlated. Moreover, formulas were calculated via linear regression to express
their relations.
- Slower moving object = more difference between head & gaze direction.
- Speed> Distance> Shape> Horizontal Moving Degree> Color.
Measuring the difference between head and gaze
orientation in virtual reality (Yuchen Qiu, 2017)

Partial Binocular Overlap in AR
● Reminder: Binocular vision and Partial Binocular Overlap
● What are the problem areas with PBO in AR?
● How bad is it?
○ What is bad and how to measure it?
● What are the optimal parameters - within possible limits?
● What can be done to alleviate the problem areas?

PBO in AR: problem areas
- Binocular Rivalry
- Situational unawareness
- Luning effect

Binocular Rivalry and Head-Worn Displays
(Patterson, 2009)
“The inhibition or suppression that binocular rivalry engenders acts upon a given
area of the retina, not upon the stimulus per se.” (Blake, 1979)

Luning
Klymenko et al. 1994 , Visual Perception in the Field-of-View of Partial Binocular Overlap Helmet-Mounted Displays (USAARL Report No. 94-40)
http://www.dtic.mil/dtic/tr/fulltext/u2/a285213.pdf

Luning
“Whereas some authors of applied studies contend that luning is less noticeable
after about 30 min of use (Grigsby & Tsou, 1994) and that it is not noticeable when
an observer engages in a demanding task (J. Melzer, personal communication April
18, 2006), there is evidence that luning increases reaction time (Klymenko,
Harding, Beasley, & Rash, 2001) and decreases detection performance (Kruk &
Longridge, 1984) for targets appearing close to the monocular flanking regions of
the partial overlap HWD.”
“To minimize the effects of luning, Grigsby and Tsou (1994) recommended a partial
overlap area of at least 40°, whereas Melzer and Moffitt (1997) and Klymenko,
Verona, Martin, Beasley, and McLean (1994) recommended the introduction of
false contour lines between the monocular and binocular regions. However,
although the introduction of false contours decreases the appearance of
fragmentation and luning, it remains to be empirically determined whether false
contours, in fact, reduce suppression.”
(Patterson, 2009)

Partial Binocular Overlap in AR
● Reminder: Binocular vision and Partial Binocular Overlap
● What are the problem areas with PBO in AR?
● How bad is it?
○ What is ‘bad’ and how to measure it?
● What are the optimal parameters - within possible limits?
● What can be done to alleviate the problem areas?

Metrics
● Performance
○ Task-dependent: speed, accuracy
● Visual fatigue
○ Pupil dilation (Murata, 2001)
○ EEG-based measuring methods
○ Self-report (questionnaires)
● Cognitive load
○ Performance in a secondary task
○ Physiology, e.g. EEG, heart-rate, pupil dilation, or even O2 intake / CO2 output comparison
○ Self-report (questionnaires)
● User behaviour patterns
○ Head / eye / body movement patterns
○ Effectively used FOV: our next experiment
● User satisfaction/frustration (questionnaire)

Measuring user comfort
We are trying carefully to apply
models of user perception from
binocular disparity research into
binocular overlap research.
Zones of comfort
(Lambooij & IJsselsteijn, 2009)

Limits of comfortable fusion (stereoscopic)
The limits of comfortable fusion decrease
with smaller, detailed, and stationary objects
and increase with larger, moving objects and
the addition of peripheral objects to the
fixation object.
Y. Y. Yeh and L. D. Silverstein, “Limits of fusion and depth judgement in
stereoscopic color displays”, Hum. Factors 32, 45–60 (1990).

Measuring visual fatigue by pupil size (Murata,
2001)

● 3 levels of FOV: 30°, 40°, 66°
● 3 levels of overlap: 30°, 40°, 66°
● 2 / 6 conditions per participant, balanced
● 40° x 80° stimulus field
● 20 x 40 items: one target (red), the rest distracters (yellow)
● Task: to look at the target and press button
● n=30
Performance metrics: our experiment I

Performance metrics: our experiment I
GLS Regression Results
=================================================================================
Dep. Variable: np.log(reaction_time) R-squared: 0.069
Model: GLS Adj. R-squared: 0.067
Method: Least Squares F-statistic: 49.61
Date: Tue, 20 Feb 2018 Prob (F-statistic): 2.46e-40
Time: 19:50:48 Log-Likelihood: -2093.1
No. Observations: 2700 AIC: 4196.
Df Residuals: 2695 BIC: 4226.
Df Model: 4
Covariance Type: nonrobust
==============================================================================
coef std err t P>|t| [0.025 0.975]
------------------------------------------------------------------------------
Intercept 0.7928 0.054 14.561 0.000 0.686 0.900
Total FOV -0.0081 0.001 -10.561 0.000 -0.010 -0.007
Binocular FOV 0.0008 0.001 0.889 0.374 -0.001 0.003
Set Order -0.1317 0.020 -6.509 0.000 -0.171 -0.092
Trial Order -0.0035 0.001 -4.542 0.000 -0.005 -0.002
==============================================================================
Omnibus: 138.684 Durbin-Watson: 1.582
Prob(Omnibus): 0.000 Jarque-Bera (JB): 159.810
Skew: 0.594 Prob(JB): 1.98e-35
Kurtosis: 3.092 Cond. No. 399.
==============================================================================
NB: Adding Participant ID to the model increases Adj.R2 to 27%

Summary
● Most things are task dependant
○ Sitting or walking
○ Requires wide FOV or precise scene perception
● Overlap size does not affect task performance, FOV does
● We need to benchmark all sorts of tasks and several relevant metrics

Our understanding so far
● In general effect similar to strong binocular disparity
● (less overlap) x (longer time) ⮞ stronger fatiguediscomfort
● (less overlap) x (closer objects) ⮞ stronger fatiguediscomfort
● (less overlap) x (depth task) ⮞ stronger fatiguediscomfort
● Narrow FOV ⮞ weak situational awareness ⮞ discomfort when
movingrotating

Partial Binocular Overlap in AR
● Reminder: Binocular vision and Partial Binocular Overlap
● What are the problems with PBO in AR?
○ Previous research and our findings so far
● How bad is it?
○ What is bad and how to measure it? Existing metrics and our ideas.
● What are the optimal parameters within possible limits?
● What can be done to alleviate the problem areas?

Partial solutions
● For binocular discrepancy
○ Nose divider (blended)
● For narrow FOV:
○ Blended outer borders
○ Smooth shape of FOV
○ Peripheral cues
● Dynamic (gaze-based) scene adjustment

Sparse Peripheral Displays
Augmenting the Field-of-View of Head-Mounted Displays with Sparse Peripheral Displays
Robert Xiao, Hrvoje Benko, Microsoft Research, 2016 https://dl.acm.org/citation.cfm?id=2858212

Glossary
● Binocular Rivalry
● Luning
● Fatigue
● Binocular Overlap
● Binocular Disparity