Eye-tracking is becoming a popular input modality for both gamers and researchers. Within gaming and general daily use, eye tracking enables options for “foveated rendering” where objects outside the fixation gaze of users are manually blurred to improve hardware performance [https://doi.org/10.1007/s41095-022-0306-4]. Social games such as VRChat [https://hello.vrchat.com/] are also examples of media spaces that were elevated into legitimate social spaces through the introduction of eye tracking and other interaction paradigms such as lip syncing and full-body motion capture [https://doi.org/10.1002/symb.629]. For researchers, eye tracking offers an open avenue for knowledge discovery around the design of user interfaces, optimizations to simulation design, and physiological studies of gaze with virtual stimuli [https://doi.org/10.1007/s10055-022-00738-z, https://doi.org/10.1007/7854_2022_409].
As a researcher interested in the design and applications of urban simulations, I am very much interested in eye tracking. However, given my background in HCI and XR, I have several concerns regarding the implementation, application, and reproducibility of eye tracking findings. These concerns give me pause on a full-throttle implementation of eye tracking into projects such as StreetSim. To alleviate these concerns, I feel compelled to do a detailed analysis of the nature of eye tracking as we currently know it, how our physiology as humans clashes and complements hardware affordances in modern VR HMDs, and whether alternatives should be considered.
This mini report will cover the following topics:
Field of View (FOV): The size of the visual field, usually in degrees, that can be viewed instantaneously [Virtual Reality: How Much Immersion Is Enough?].
Field of Regard (FOR): The total size of the visual field surrounding a person, in degrees [Virtual Reality: How Much Immersion Is Enough?]. This is distinct from the FOV, in the sense that a VR environment has a full $360\degree$FOR while a CAVE system might have a $270\degree$ FOR if they don’t have a projector screen behind the user [The Immersive Virtual Environment of the digital fulldome: Considerations of relevant psychological processes]. We don’t typically adress this in VR literature because HMDs are naturally $360\degree$ FOR and won’t really be consequential for our situation.
Steradian or Solid Angle: A 3D representation of an angle within a sphere. In contrast to flat angles that are defined in 2D space, a steradian is used to represent a conic area within a sphere from its center. Steradians are defined by $\Omega$ in terms of the sphere radius $r$ and circular area $A$:
$$ \Omega=\frac{A}{r^2} $$
Vestibular-Ocular Reflex (VOR): A phenomenon where, during a gaze fixation where both the head and eyes rotate, the eye will remain fixated on target while the head continues to rotate [Types of Eye Movements and Their Functions] [Eye-head coordination during large gaze shifts, pg. 768]. This motion is intended to, by the end of the total movement, to have the eyes facing the relative forward direction of the head and thus have the eyes rest at a comfortable, neutral state while the head remains fixated on target.
Range of Motion (ROM): a quantitative way to describe the motion of a joint or body part. Specific literature will refer to these in 3D, dictating 3 specific planes of reference to look at: 1) “Sagittal”, 2) “Frontal”, and 3) “Horizontal” or “Transverse”.
https://physiquedevelopment.com/planes-of-motion-sagittal-frontal-transverse/
Note that I am not a specialist on the specific biology of humans and cannot go too deep into this topic.
According to “The Limits of Human Vision” by Michael F. Deering (1998, pg. 5), “The FOV of a single human eye is about one third of the entire 4π steradians FOV.” For the eyes, each eyeball represents $\frac{4\pi}{3}$ steradians. This converts to approximately $141.0575587\degree$.
According to [https://doi.org/10.1145/3361218] (and [https://doi.org/10.1093/acprof:oso/9780198570943.001.0001], though I can’t access this report directly), healthy adult eyes have a motion range of $50\degree$ in any direction; the total FOV of human vision is $210\degree$ horizontally and $120\degree$ vertically ($50\degree$ upward, $70\degree$ downward).
In [Ch. 1, pg. 6 of an older report from 1938], the authors report that the average individuals’ eye individually has a range of motion of $60\degree$ upward, $60\degree$ inward, $75\degree$ downward, $100\degree$ outward, and between $45\degree$ and $50\degree$ in the direction of the nose. This would be roughly equal to $145\degree$ to $150\degree$ horizontally, assuming we use the toward-the-nose direction as the inward movement. Vertically, we have a total of $135\degree$. In binocular vision, the overlap between the two eyes is nearly circular with a diameter of approx. $120\degree$. binocular vision in total covers a range of $200\degree$ horizontal and $130\degree$ vertical. This 1938 report also emphasizes some particular ranges within this visual field:
Here’s two screenshots of the visual field of the eyes (left = monocular, single eye; right = binocular vision):
