Reducing disparity eases the conflict at the cost of flattening space

What it is

Scaling down the horizontal difference between the two eye images, or narrowing the virtual camera baseline, is disparity reduction (stereo scaling). Vergence demand shrinks with it, so the fight against the headset’s fixed focus eases a notch and the eyes tire less. The same move flattens the scene in depth: a metre of front-to-back becomes a few centimetres of relief. Cinema 3D does this not to be “more 3D”, but to keep people in the seat.

It is a trade made after the conflict is already granted, not another pass over comfort distance.

Why it happens

Headset focus is nailed to a screen; vergence follows disparity. Large disparity drives the eyes inward while accommodation stays on the screen — that is the geometry of the conflict. Multiply disparity by a factor below one and the vergence angle is pulled back toward the screen plane; the amplitude of the conflict falls. Depth perception uses that same disparity, so recovered depth differences thin by the same factor. Occlusion and perspective remain, so the world does not become a pure sticker, but thickness you could measure with a hand is lost.

Another recipe squeezes both near and far toward a chosen depth plane, often near a comfortable vergence. Pop-out and sink-in are shaved off; the middle slab remains. That hurts the subject less than uniform scaling, and it also collapses interface layers that were separated by a large depth gap.

The trade is monotonic: gain near 0 is 2D; gain near 1 brings the conflict back intact. There is no third setting that is both thick and effortless.

Studying it

Sweep disparity gain (0 to 1, sometimes a little above) in one scene, and measure comfort and a depth task together. Comfort: in-session verbal ratings or post-session ocular-symptom items. Depth: ordering, grasp landing, relative-distance estimates.

Independent variables: disparity gain, depth range of the content, viewing versus manipulation as the task. Dependent variables: discomfort ratings, compression factor on depth estimates, grasp error along the depth axis.

Do not fold cinematic long shots and near-field manipulation into one mean — the former tolerate heavy compression; the latter start missing grasps below some gain.

Where it stops holding

Watch-only content (a film, a distant gallery) can run gain very low; viewers may even prefer a mild 3D. Tasks that must align, twist or stack within an arm’s length will, at the same compression, glue front and back into one layer. Stereo-anomalous people barely eat disparity gain, so turning it down or up changes little; their comfort moves with other things. Gain above 1 (hyperstereo) enlarges both the conflict and size distortion, and should not be the default “more immersion” switch. A twenty-minute lab comfort curve does not extrapolate to two hours of near-field work.

Applying it

• Give disparity a user-facing gain, defaulting a little below geometric correctness; let near-field tools sit on a higher step and far scenery on a lower one.
• Do not put “stereo at maximum” at the top of a quality menu — that top step is often the most tiring.
• After reducing disparity, re-check grasps and panel ordering. If front and back start being mis-clicked, restore layering (occlusion, spacing); do not quietly crank gain back up.
• How to check: run the same manipulation at gain 1.0, 0.5 and 0.0. The step where discomfort falls and depth error rises is the trade in action. Pick the step by whether the task is watching or handling, not by a slogan about “stereoscopic feel”.

Related

• Same group: N1.10.1 Binocular disparity only provides useful depth at near distances · N1.10.2 A wrong rendered IPD globally changes size judgments · N1.10.3 Some people have no stereo vision and need monocular depth-cue fallback
• Nearby: N1.04 Vergence–Accommodation Conflict · N3.02 Interface Distance
• Search terms: disparity reduction · stereo scaling · depth compression