Depth Perception

Question 1

Stereopis

Answer 1

Retinal disparity

Question 2

What is retinal disparity

Answer 2

Our eyes have two different views of the world
Any object not on the horopter has a retinal display
If we can calculate the retinal disparity we know how far the object is away from our ‘horopter’
If we can create images that have retinal disparity they should appear to have depth

Question 3

Horopter

Answer 3

The locus of points in space that have the same disparity as fixation

Question 4

Ways of getting different images into each eye

Answer 4

Red - green anaglyphs
Polaroid over each eye
Mirrors
Free fuse
Autostereograms (magic eye)

Question 5

What is polarity

Answer 5

Disparity governs whether seen as in front of behind horopter

Amount of disparity governs the amount depth seen

Question 6

Random dot stereograms

Answer 6

Bella julesz 1961
- pattern of random dots is copied with some of the dots shifted

Each pattern presented to diffferent eye
Each dot alone sees nothing but random dots

Together the eyes extract the retinal disparity and see form

Question 7

Depth

Answer 7

Depth must precede form

Question 8

Autostereograms

Answer 8

2D images designed to reveal a hidden 3D object by forcing the viewers eyes to diverge or converge, bypassing normal focus.

Repeating patterns horizontally, brain fuses different parts of the image, perceiving depth based on pixel shifts

Eyes trick the brain into seeing 3D separating signals from each eye, causing it to see a hidden object at different depth background

Question 9

Neural basis of disparity sensitivity

Answer 9

Cells cannot be driven by both eyes at the level of the retina

Left and right eyes drive different layers of the LGN - no binocular cells

V1 where first place information from two eyes come together

Many cells are disparity sensitive = Barlow et al 1968

Question 10

Stereoblindness

Answer 10

Many people (10%) have problems with thier stereo vision

• if there are early problems then the brain doesn’t learn to put information from the two eyes together and this cue to depth is lost

Such people sometimes dont know that they have this problem - must be other cues to depth

Question 11

Motion parallax

Answer 11

Movement through the world, eyes cause the image on our retina to move

Thing close to us move faster so if we have 2D dots and move them at different speeeds they can look like they have depth

Question 12

Rostering snake illusion

Answer 12

High contrast dark blue to light
Low contrast green to black

High contrast is processed quicker than low contrast

Question 13

Movement of dots study

Answer 13

Rodger’s and Graham 1979 tolled the movement of dots to the movement of the head

• subject ‘saw’ static 3-D surface when looking at moving 2-D surface

Question 14

Motion after effect (waterfall illusion)

Answer 14

Prolonged movement in one direction leads to a stationary image appearing to move in opposite direction

Watch waterfall and then look at the stone river and the stone start shooting up

Bank of motion detectors

Question 15

Pictorial cues to depth

Answer 15

Despite a lack of stereo cues and lack of motion we can still see depth in pictures

• we use a lot of cues such as:
  Interposition, height, size, perspective. Shadows

Question 16

Tilt effect

Answer 16

Must be a place code for motion direction,

Left go right
Down go up

Question 17

Apparent motion

Answer 17

TV, flick books, but timing and spacing are all important to produce motion

Still frames that move causing the motion cue , see movie frames at a high level

Question 18

Size

Answer 18

For a given object size will cast a smaller and smaller retinal image as it gets further away

Question 19

Texture radients

Question 20

Kinematograms study

Answer 20

Braddick 1974, 1980
— subjects tried to identify the orientation of moving area of dots

— only possible with small displacement (<0.25), small time intervals (<100ms)

Easy to identify whether thier is a small vertical or horizontal change

Question 21

Viewing techniques

Answer 21

Divergent (parallel) viewing
- standard technique, called looking through imaging, requires focusing on a point behind the surface

Convergent (cross-eyed) viewing
- crossing eyes to look closer than the image, used for cross-view specific images

Question 22

Stereoacuity connection

Answer 22

Self-reported ability to see Magic eye images is highly predictive of an indivdual’s steroacuity (ability to perceive depth and relative distance using both eyes, binocular vision, often expressed in seconds, precision of the 3D depth

Question 24

Pictorial cues to depth

Answer 24

Lack of stereo cues and lack of motion we can still see depth in pictures

Use a lot of cues;
Interposition, height, size, perspective, shadows

Question 25

Texture gradients

Answer 25

TG - gradual change in appearance of textures in appearance of textures on surfaces, provides information about slant and distance Density - surface extends into teh distance texture elements become more densely packed, Clarity - fine details on nearby objects are seen clearly, whereas such details are less visible in the distance Foreshortening - surface tits away from observer, circular texture elements may appear as ovals (foreshortened), indicating change in slant Application - gradients used to interpret the depth of surfaces such as sandy beaches, brick walled, road surfaces

Question 26

Shape and shading

Answer 26

Vital cues to 3D shapes and orientation fo objects - shading from shading = luminance depend on surface orientation relative to light source, allowing visual system to reconstruct 3D shapes Light assumption - sum light source originate from above, brighter top surface and darker bottom surface typically interpreted as a mound, reverse indicates a crater Shading and texture cooperate to define surface geometry, shading often plays a stronger role in identifying complex 3D shapes.

Question 27

Light from above assumption

Answer 27

Ramachandrans 1988 - shape form shading demonstrates taht human visual system relies on specific, built in assumption to perceive depth. Strong, default assumption that plight comes form above (top-left), shading gradients used to interpret surfaces convex its or concavity Global vs local operation = complex scenes, brain tends to assume a single, consistent light source for the entire scene Pre-attentive process = shape-from-shading processed rapidly and pre-attentively meaning depth is perceived before conscious attention paid to the object Revolutionaised monocular cues

Question 28

Cast shadows

Answer 28

Moving cast shadows are a powerful, dominant, cue for perceiving 3D motion of objects Manipulating the motion of a cast shawls can change the perceived 3D trajectory of the object casting Stationary light source constraint - human visual system tends to assume fixed, interpreting shadow movement as object movement rather than light source movement Shadowing darkness 0 contrast and darkness of a shadow help define it, unnatural, light shadows can induce strong illusory motion in depth

Question 29

Distortions

Answer 29

Misplaced depth, Understands depth is used to then decide on how large or small things are, their spatial relationships Emmert’s law - images of same retinal size will appear different sizes when located at different distances (linear relationships)

Question 30

Emmerts law

Answer 30

Perceived size of a constant retail image (like an afterimage) increases as its perceived distance from the observer increases Optic flow - pattern of light changes on the retina as an observer moves through an environment, providing cues direction and speed via a ‘focus of expansion’ True motion calculation - Brian distinguished between a single moving object and multiple stationary objects within a scene

Question 31

Size constancy

Answer 31

Perceptual ability to maintain a stable percetion of an objects size despite changes in its retinal image size due to distance Emmerts law quantifies this by stating than an objects perceived linear size is proportional to its perceived distance Allows you to perceive an object as the same size regardless of whether it is close or far away If retinal image is constant then as the perceived distance increases the perceived size must also increase

Question 32

Misplaced depth

Answer 32

Distortions occurs due to misplaced depth Hollow face = top-down processing (knowledge overriding sensory input), brain assumes face as convex ponzu illusion = two identical lines appear different sizes, based on linear perspective, Brian interprets upper line as more distant Muller-layer illusion = two equal lines appear different lengths due to arrow like ends, inward short, outward long, interpreted as depth cues, brain applies size constancy scaling Ames room - distorted room makes people appear giant/ tiny, irregularly shaped, Brian assumes that room is rectangular, misinterprets depth as misjudges size

Question 33

Illusions

Answer 33

All show that perception is not a direct copy of reality but rather an interpretation based on depth cues (perspective, shading, size) and assumptions