Jump to content

Parallax

From VR & AR Wiki

Parallax is the apparent shift in the position of an object against its background when the viewpoint changes. Objects at different distances shift by different amounts: a nearer object appears to move more than a farther one for the same change in viewpoint.[1] In virtual reality and augmented reality, parallax is one of the main cues the visual system uses to judge distance, and reproducing it correctly is a big part of why a headset feels three-dimensional rather than flat.

In the context of headsets the term covers two related effects that have different sources. Binocular parallax is the difference between the two eyes' views at a single instant, and it is what Stereoscopic 3D rendering exploits. Motion parallax is the change in the scene as the viewpoint itself moves, for example when the user moves their head or body, and it requires the headset to know where the head is in space. The stub on this page originally described motion parallax alone, which is the form most people picture when they hear the word, but a headset deals with both.

Binocular parallax

The two eyes sit a few centimetres apart, so each one sees the world from a slightly different position and the two retinal images do not match exactly. This small difference is called binocular disparity, and the horizontal component of it is sometimes called binocular parallax.[2] The amount of disparity depends on how far away a feature is, so the brain can use it to recover depth through a process called stereopsis.[2] Of the binocular cues, disparity is the strongest, and it works even when the observer and the scene are completely still.[3]

Stereoscopic headsets recreate binocular parallax directly. The system renders a separate image for each eye from two virtual viewpoints offset by roughly the user's interpupillary distance, and a lens in front of each display sends the correct image to the correct eye. The average adult interpupillary distance is about 63 mm, with most people falling between 50 and 75 mm, which is why many headsets let the user adjust the lens spacing.[4] Because the two images differ in the same way two real eyes' images would, the brain fuses them and reads depth from the disparity. The idea that horizontal disparity alone is enough to produce a sense of depth goes back to Charles Wheatstone, who demonstrated it with his stereoscope in 1838.[2]

Motion parallax

Motion parallax is the depth information that appears when the viewpoint moves through a scene. As an observer moves, objects at different distances sweep across the field of view at different speeds: nearby objects appear to move quickly, while distant objects appear to move slowly or barely at all.[5] The familiar example is looking out of the side window of a moving car, where roadside objects rush past while hills on the horizon hardly seem to shift.[5] Unlike binocular parallax, motion parallax is a monocular cue, so it works with a single eye and does not depend on combining two retinal images.[5]

In a headset, motion parallax only appears if the system tracks the position of the head, not just its orientation. A headset with positional tracking has six degrees of freedom, or 6DoF, meaning it tracks the three rotations of the head plus translation along the three spatial axes (left and right, up and down, forward and back).[6] When the user leans in to look at something or peers around an edge, a 6DoF headset moves the virtual viewpoint to match, and nearby objects shift more than far ones exactly as they would in the real world.[6]

A headset with only three degrees of freedom, or 3DoF, tracks rotation but not position. It knows which way the head is pointing but not where it has moved, so leaning or stepping does not change the rendered viewpoint.[6] For this reason a 3DoF headset cannot deliver true motion parallax; the scene rotates around the user but does not open up as they move into it. This is one of the clearest practical differences between simple 3DoF viewers and full 6DoF systems.[7]

Why correct parallax matters

Parallax is one of several cues the brain combines to build a sense of depth perception, and in VR it carries a lot of weight. Some writers argue that motion parallax matters more for the feeling of being inside a 3D space than stereoscopy does, because it responds to the user's own movement.[6] Getting it right also affects comfort. When a 6DoF headset reproduces motion parallax, the visual scene moves in a way that agrees with what the inner ear reports about head movement, which helps reduce the sensory mismatch that causes motion sickness.[6][7]

Binocular parallax brings a complication of its own. In most current headsets the displays sit at a fixed optical distance, so the eyes always focus, or accommodate, at that one plane even though the rendered binocular parallax tells them to converge at the depth of whatever they are looking at. When those two distances disagree, the result is the vergence-accommodation conflict, which can cause eye strain, fatigue, and headaches during longer sessions.[8] Varifocal and other experimental display designs try to move the focal plane to match the convergence distance and close the gap.[8]

This is also why ocular parallax, the shift caused by rotation of the eye within its socket, is a separate effect and should not be confused with the head and body movement that drives motion parallax.

See also

References