Presets
Playback
Geometry
Numerical checks
The science
What is and is not modelled. The title this page replaces promised something about consciousness and "hyper-space". Nothing like that is simulated here. What is real and measurable is how two eyes (or two cameras) turn distance into a small angle called binocular disparity, and how finely that angle can be used to judge depth.
Two eyes at (±b/2, 0) look at a fixation point F = (0, D). A point P lands on the two retinas at slightly different places. Relative disparity is the difference of the two retinal eccentricities, which equals the difference of the vergence angles (the angle between the two lines of sight):
Depth resolution and range
If the smallest usable disparity is δmin (stereoacuity), the smallest detectable depth step at distance z is Δz = δmin z² / b (first-order exact form δmin(z² + b²/4)/b). It grows with the square of distance and shrinks with a wider baseline b, which is why survey cameras and rangefinders use long baselines. Beyond roughly b / δmin there is no usable disparity at all.
Points with zero disparity lie on the Vieth-Müller circle through both eyes and the fixation point (inscribed-angle theorem). That is a geometric circle; the measured human horopter differs from it.
Checks shown on the page
- Disparity by projecting the object onto two pinhole retinas vs disparity from the angle between the lines of sight (different algorithm, same quantity), including any sideways offset.
- Zero disparity (to numerical precision) for nine points on the Vieth-Müller circle.
- Smallest depth step: bisection on the exact geometry vs the first-order formula.
- Monte Carlo of a noisy disparity measurement (20 000 draws, noise = stereoacuity): spread of the distance estimate vs
Δzpropagated through the formula.
Honest limits
- Horizontal plane only; no vertical disparity, torsion, eye-model optics or cyclopean fusion limits (Panum's area). Real depth perception also uses motion parallax, size, shading, occlusion and more, and its stereo part is learned and varies a lot between people; some people have no useful stereo vision.
- Stereoacuity (typically tens of arcseconds under good lab conditions) and eye separation (about 63 mm for adults) are typical values, not constants: check them against a vision-science source before quoting.
- Marker poles are drawn larger with distance so they stay visible; geometry (positions, angles) is not scaled.
The 2D views
The plan view colours every point of the horizontal plane by its disparity relative to the fixation point: orange is nearer (crossed), blue is farther (uncrossed), grey is closer to zero than the stereoacuity, so it cannot be told from the fixation plane. The grey band around the violet Vieth-Müller circle is the depth step made visible; it widens with distance and shrinks with a larger baseline. Below the map the two retinal images show where the object lands relative to each fovea; their difference is the disparity, drawn on a magnified gauge. New checks: the zero contour of the map found by bisection lies exactly on the analytic circle, and the midline graph agrees with the closed form.