22.1.

2018 | CS3VR16 – Virtual Reality Lecture 3

Virtual Reality – visual performance characteristics

 Smallest resolvable separation in a pattern ~0.0083°
 Field of view
 Limitations
o Blind spot
o Cannot see from behind
 What we see
o Limited sensory organs
o Colour: sum of narrow bandwidth ‘notches’ (of
electromagnetic spectrum)
o Misses most of spectrum
o Discrete images merge into continuous motion
o Hardly measure depth (depth-perception; worked out by
processing)
o Interpret/assumed vision
o Optical illusions show limitations
o Useful to exploit in VR design
 Psychology of visual system
o Visual perception: colour, form, pattern recognition, motion,
depth
o Depth/motion more difficult
 No receptors for depth
 Inferred
 Prone to error
o Link between depth/motion
  Psychology of position
o Work out object position relative to observer (direction +
distance)
o Direction (easy): 2D retina  brain  2D angular coordinates
o Depth (harder): no absolute method
o Books on topic; focus on depth perception (impression of
depth)
o Ames room – used in Lord of the Ring films
o Depth cues
 Used to work out how far things are
 Oculomotor almost a metric
o Oculomotor
 Use muscles in eye as measure
 Accommodation – eye focus effort required can
describe depth
 Convergence – orientation of eyes (better metric)
o Static, monocular
 Interposition
 Partial obscuring
 Simple, strong, overrides most other methods
 Size – expected depth affects perceived size
 Linear perspective/texture
 Shading – includes lighting/shadowing
 Atmospheric – distant objects less clear, more tinted
(blue hue usually adopted by artists)
 Shadows – interpret relative positioning
o Stereopsis (seeing solid) – retinal disparity
 Each eye slightly different view

2
  Brain merges to get solid (3D) image
o Dynamic
 Motion parallax
 When objects/viewer move, closer objects move
faster across retina
 Requires motion for effect
 Kinetic depth
 Objects can seem 3D, 2D when stationary
 Brain tries to make sense of word from
limited/incomplete sensory input
 Motion perception
o Real world: continuous motion; VR/films/senses/etc.: discrete
motion
o If successive stimuli close, continuous motion perceived
(sufficiently often to convince senses)
o Motion is primary quality (employs multiple senses)
o Different threshold times
 Vision (flicker fusion) ~17ms*
 Audition (flutter fusion) ~10ms*
 Touch (vibration) ~5ms*
o Note: fusion not necessary for motion to occur, but
preferable
o Apparent motion
 Lots of varieties: short rang, long range,
transformational, biological
 Discreteness has other effects
 ‘Wagon wheel’ effect: after sufficient speed,
wheel seems to move backwards
 Low frequency v. high frequency perception

3
  Visual display technology
o Desktop, wearable, projection
o Most 3D cues discussed part of 2D image
o Stereoscopy needs dedicated hardware
 Stereoscopic display technology
o Each eye sees a different image
o If same display surface, selectively block or transmit
 Interlaced display
 Glasses s each eye sees own frame
o Most displays have ghosting/crosstalk  discomfort, loss of
presence
o Varieties
 Active – liquid crystal shutter glasses
 Passive
 Polarisation filtering; each lens allows only certain
directions of light
 Linear polarisation
o Frame vertical/horizontal
o Cheaper/easier to make
o High percentage light
absorption/transmission
o User can’t rotate head
 Circular polarisation
o Clockwise/anti-clockwise frame
o Can have more ghosting
o Invariant of viewer orientation
o Used in modern cinemas
 Frequency multiplexing

4
 o Old style anaglyph
o Block different parts of spectrum
o Invariant of head rotation
o Loses colour info
o Infinitec preserve colour
o Any display surface useable
 Autostereoscopic display
 No glasses
 Lens/barriers so each eye sees its part of display
 Fixed barriers
 Limited viewing angle + ‘sweet spots’
 Can allow multiple viewing position/less
resolution
 Doesn’t scale well
 Parallax barriers lenticular lens
 Movable barriers
 More viewing angles
 Track user eye position
 Adjust barrier to match
 Get ‘look around’ effect
 One viewer at a time
 Comparing visual displays
o Criteria for selecting
o Tech
 Field of view
 Resolution
 Update rate (pass flicker fusion)

5
  Interface w/ tracking methods
 Associability with other sense displays (ex:
audio/haptic)
o Ergonomic
 User mobility
 Environment requirements
 Throughput (ow many can experience VR together)
 Safety
 Time
 Encumbrance
o Others
 Portability
 Cost
 Desktop displays
o Cheap/common/accessible
o Small field of view
o Use all stereo
o Techniques discussed
 Head mounted displays
o Boom mounted/hand held
o Expensive (used to be)
o OculusRift/Google Cardboard
o Separate display for each eye
o Optics magnify image, change focal length
o Generally low resolution
o Usually limited field of view
o Real world normally blocked from view
o Weight + comfort big issue (glasses)

6
 o Latency tracking
o Can’t see own body
o Balance/movement affected
o Unnatural body positions
 Coming 2018
o Oculus Go
o Oculus Santa Cruz (with hand wearables)
o Magic Leap One (glasses lite)
o Neurable’s brain-scanning headband (think it to happen)
o More AR (adding to real world)
 Projection displays
o 1+ projectors/screen
o Large area, good field of view
o 2D stereo
o Usually rear projected (prevent shadows of user distorting the
view)
o User can see own body
o Multi-user
o Vision Dome
o Spheres cubic room
 Computer assisted virtual environment (CAVE)
o 3-6 large screens for visual immersion
o Usually 1+ powerful graphics engines
o Can pull external equipment in
o Allows multiple people
o Visbox comparison (CAVE) to Oculus Rift
 Projection display issues

7
 o Geometric continuity (aligned imaging) + photometric
continuity (consistent resolution)
o Wall displays
 Multi projection
 Continuity important
 Tiling/blending
 Next week: audio perception
Graphics
 3D scene representation
o Representations – points, lines/line segments, polygons,
surfaces, solids, voxel (3D pixels), etc.
o Require
 Points/lines/surfaces on objects
 Vectors (between points)/rays (of light)/ vector + matrix
operations
o Coordinate systems
 Geometric base to form computations: coordinate
systems
 Cartesian rectilinear (x,y,z)
o Left hand/right hand rule
o To convert between two, invert z axis
 Spherical/polar/angular (r,_,_) – for 2D, x = rcosƟ,
y= rsinƟ
 3D primitives
o Point
 Specifies a location (no volume)
𝑥
 3 values: p = (x,y,z) /[𝑦]
𝑧
o Vector

8
  Direction + magnitude
𝑑𝑥
 𝑣̅ = (𝑑𝑥, 𝑑𝑦, 𝑑𝑧)/ [𝑑𝑦]
𝑑𝑧
o Operations: adding/subtraction (must be the same size);
multiply/divide (by scalar)
o Vector length + normalisation
 Magnitude  ‖𝑣̅ ‖ = √𝑥 2 + 𝑦 2 + 𝑧 2
𝑣̅
 Normalised when ‖𝑣̅ ‖ = 1 or 𝑣̅ = ‖𝑣̅‖

o Dot product
 𝑎̅ = (𝑎1 , 𝑎2 , 𝑎3 ), 𝑏̅ = (𝑏1 , 𝑏2 , 𝑏3 )
 𝑎 ⋅ 𝑏 results in scalar value
o Angle between 2 vectors
 Dot product also defined as 𝑎̅ ⋅ 𝑏̅ = ‖𝑎̅‖‖𝑏̅‖ 𝑐𝑜𝑠 𝜃
𝑎̅⋅𝑏̅
 Hence 𝑐𝑜𝑠 𝜃 = ‖𝑎̅‖‖𝑏̅‖

 Ɵ found by taking inverse cos-1()

 2 vectors orthogonal if 𝑎̅ ⋅ 𝑏̅ = 0
 3D line
o Point 𝑝̅ w/ direction 𝑣̅ specifies line
o ̅̅̅
𝑝1 point on same line ̅̅̅
𝑝1 = 𝑝̅ + 𝑡𝑣̅
o T is scalar (-  t  )
o If 0  t  , figure is a ray
 3D line segment
o 2 points : ̅̅̅
𝑝3 = 𝑝̅ + 𝑡(𝑝 𝑝1 where 0  t  1
̅̅̅2 − ̅̅̅)
o Confirms ̅̅̅
𝑝2 − ̅̅̅
𝑝1 = 𝑣̅
 Matrices
o 𝑚×𝑚

9
 𝑎 𝑏
o Determinant |𝑀| = [ ] = 𝑎𝑑 − 𝑏𝑐
𝑐 𝑑
o Useful for vector cross product (3x3 usually): 𝑎̅ × 𝑏̅ =
‖𝑎̅‖‖𝑏̅‖𝑛̂ sin 𝜃 (where n is normal vector; returns a vector)
o Cross product used for normal of 2 vectors
 3D plane
o Represents surfaces mathematical plane
o Defined by
 Pint + normal vector to plane
 Any 3 points not co-linear
 Bound plane/polygon
o Triangle, quadrilateral, convex, star concave, self-intersecting
o If hole required, more than one polygon
o Polygon requires straight edges, all in one plane (no curves)
 Matrix multiplication
o Columns of first (A) matrix must be equal to rows of second
(B)
o 4x4 matrix multiplication used a lot in VR

10