Field of view shapes immersion and peripheral awareness
Aliases: FOV · HMD FOV · visual field of the display · surrounding
What it is
The eyes cover about two hundred degrees horizontally; a headset typically lights ninety to a hundred and twenty of them. The remaining ring is still the real room, the nose pads, the frame. Field of view (FOV) does two things at once: how much of vision the display occupies — the “surrounding” item in immersion as a system property — and whether lateral motion, an approaching person, a flash, can still be registered without a head turn — peripheral awareness. Field of view is not image quality. It is how much retina is occupied by the virtual world rather than by a real bezel.
Optical FOV, rendered FOV, and binocular overlap are often three different numbers. The figure written into a spec has to say which one was measured, or “a hundred and twenty degrees” cannot be compared.
Why it happens
The surrounding dimension of immersion feeds directly on field of view: the farther the display pushes the real bezel, the more evidence there is for reading the input as an environment rather than as a screen in front of the face. That quantity can be changed by the system without a questionnaire. Peripheral awareness takes a different path. Peripheral retina is sensitive to motion, luminance transients, and looming, not to detail. Narrow the field and the sampling window for those alarms narrows with it. People are not “unable to read type at the side”; the side is not lit by the virtual world, so an alarm has to wait until it is turned into the fovea.
The head can compensate. A turn feeds a new patch of world into the field, so a small field does not mean the world is absent. The cost is time and posture: every act of noticing becomes a deliberate scan, and the central task is interrupted while scanning. Field of view governs what can be known without turning, not what can be known after enough turns.
Studying it
Probe peripheral awareness with a dual task: a continuous job in the centre (reading values, aiming) and a motion or figure flashed at varying eccentricity, recording detection and whether a head turn occurred. On the immersion side, measure surrounding objectively: solid angle of the display in the visual field, whether the real bezel is visible — not “did you feel immersed.”
Independent variables: horizontal and vertical field of view, binocular overlap, eccentricity and motion type of the probe. Dependent variables: detection without a head turn, latency to the first turn, interruptions of the central task; for surrounding, reports of a visible bezel or a camera shot of light leak around the eyes.
When comparing headsets, split optical field from software crop. Software that crops rendering to eighty degrees on hardware that still has a hundred is measuring the crop, not the lens.
Where it stops holding
Optical see-through leaves the real periphery in place, so a narrower field mainly hurts surrounding for virtual content and damages real lateral alarms less than video see-through, which turns unrendered angles into a black frame or a blurry pass-through. Vertical field is more critical than horizontal for desk work, looking at the hands, looking at the feet; quoting only the horizontal underestimates the loss for seated work. Foveated rendering lowers peripheral resolution while keeping solid angle; detail awareness drops and motion awareness can survive — do not collapse both into “the field got smaller.” Some people habitually scan with the head, and a lab “no-turn detection rate” will under-read their awareness in real use; count the cost of the turn into task time rather than reporting misses alone.
Applying it
- Treat field of view as a budget for surrounding and awareness, not as a marketing number. Events that must be registered without a head turn (someone entering from the side, a tool coming into view, a status lamp) belong inside the eccentricity the current device covers without turning. Confirm on that device; do not lay out to the eye’s two hundred degrees.
- Write down whether acceptance measured optical FOV, rendered FOV, or overlap; list all three. Do not advertise only the largest.
- For a heavy central task, either give lateral events an entry that the field actually covers, or accept that they require a head turn — and write those few hundred milliseconds into the time budget.
- How to check: put an unpausable job in the centre and send a moving target in at thirty, fifty, and seventy degrees left and right. Record detection without a turn. The angle at which the curve falls off on this device is the working line for peripheral awareness, not the anatomical line of the eye.