Display Systems Will Steptoe (Based on slides by Anthony Steed)
Overview • Critiques of Display Technologies – Colour Gamut – Brightness & Contrast – Frame Rate • Depth Cues – Virtualisation – Cameras
Human-Perceptible Colour Gamut • Colour is the perceptible proportion of the electromagnetic spectrum • International Commission on Illumination (CIE) Chromaticity Diagram (1931). Describes ‘average’ properties of human eye. • Curve is monochromatic (single-wavelength, nm) light International Commission on Illumination (CIE) Chromaticity Diagram (1931)
Standard Computer Display Gamut • Only a portion of the human colour gamut is reproduced by typical computer displays. • Image shows typical CRT/LCD display. Different devices have different gamuts: type (printer, projector, HDR display...), manufacturer, model. • Image on previous slide describes colours outside the sRGB gamut, so depending on the display calibration, they may not be displayed properly!
Primary Colours for Computer Displays • Emission spectra of CRT displays • Displays mix light instead of colour: red, blue, green (RGB) • Light is additive rather than CRT Phosphor Spectrogram subtractive as used by artists • Additive colour primaries are secondary colours of subtractive and vice versa. Additive (displays) Subtractive (paint)
Brightness & Contrast • Brightness – Projectors 2000 lumens (power of light + human eye) – Screen 500 candelas/m 2 (intensity + direction + eye) – Luminance may be different for different colours • Contrast Ratio – Ratio between black level and white – 1000:1 is good – Very difficult to measure accurately – Also depends on response time (time to change between any two levels). Static vs Dynamic ratio.
Frame Rate • Image sources are various rates – Film at 24Hz/96Hz – PAL TV at 50Hz – NTSC TV at 59.97Hz • Screens – CRTS have typically matched TV – Standard LCD panels up to 75Hz – LCD now up to120Hz (double NTSC/HDTV for stereo) • Projectors – CRT 120Hz – DLP only recently matching this (digital cinema rear projection)
Depth Cues • Physiological – Body has to do something in response to “reality” of the 3D world: doing that thing is a depth cue itself • Psychological – Brain perceives some characteristics of a visual scene and infers its 3D nature via empirical experience (Baysian inference)
Three Levels of Virtualisation • Virtual Space – see a 3D object when looking at something inscribed on a flat sheet – Perspective cues, lighting, shading • Virtual Image – perception of an object with depth – Stereo disparity • Virtual Environment – perception of a surrounding environment – head-slaved parallax, accommodation and vergence, surrounding display (describes CAVE)
Level 1: Virtual Space Half-Life 2, Valve
Level 1: Virtual Space Shadow of the Colossus, Sony
Level 2: Stereo Pairs First implementation of stereoscopy in 1840 by presenting two slightly different images to each eye Keystone View Company - Ruins of the Granite Temple, the Sphinx and the Great Pyramid (Oliver Wendell Holmes Stereoscopic Research Library)
Level 2: Virtual Image - Combines multiple cues: - stereopsis, accomodation, and vergence - Actual perceptual scale of space is not arbitrary
Level 3: Virtual Environment • Co-ordinated multisensory display with observer-slaved motion parallax (head tracking) • Depth-of-focus variation and wide field of view without visible restriction in field-of-view, postural reflex, consistent near-reflex, vestibular ocular-reflex • Requires (almost) full immersion because of the requirements for head-related and egocentric display • User can actively explore the (visual) environment by moving naturally
Psychological Depth Cues • Linear perspective • Shading • Shadows • Aerial Perspective • Occlusion • Texture gradient • Fogging
How do we see in stereo? • Inter-Pupilary Distance (IPD) or binocular disparity ~ 65mm • Each eye has different view of same object • Perceptual fusion of two views 30cm (cyclopean union), subsequent perception of depth • Inputs from 2 eyes converge on the same cortical neurons in 6.5cm visual cortex (V1) • Calculation of how different the 2 views are leads to model of depth
Left eye Right eye Binocular percept Adapted from [Purves & Lotto]
Physiological Depth Cues • Accommodation – Focal length of eyes adjust to focus at different points in scene. – Change thickness of lens: relax or tense ciliary muscles. • Convergence – Rotation of eyes inwards: view near objects – Rotation of eyes outwards (~parallel): view far objects – Powerful cue Adapted from [Slater, Steed, Chrysanthou]
Accommodation and Convergence • Usually work in conjunction with each other • This correspondence is not physiologically determined • Learned by experience • Is broken when looking at eg screen based stereo views
Autostereograms Override vergence by forcing your eyes parallel
Level of Definition Cues Technology virtualization ALL PSYCHOLOGICAL CUES: Virtual Pen & Paper Linear persp., 3D objects Space + Perspective… Shading, inscribed on a ?? Shadows, flat sheet Aerial persp., Occlusion, Texture cues PHYSIOLOGICAL “Stereoscope” CUES Perception of 1830s Virtual Stereoscopic objects with Charles Wheatstone Image disparity, depth Accomodation* Stereo screens… Convergence* Objects All cues Cave, HMDs… Virtual slaved motion consistent with ~1980s - ‘90s Environment observer motion parallax
Cameras: Stereo Graphics Creation • Measure positions of the left and right eye • For each display surface render the image, with the field of view that matches the eye’s view of that surface • Easy in a CAVE – Image remains flat • More difficult in head-mounted displays – Image often distorted by the optics, no longer square when seen.
Setting up a stereo view • Project different views of the Image plane same scene to each eye at the ~same time Left Eye • If want to present point as R1 being in front of screen: L2 (virtual point P1) use L1, R1 P1 P2 R2 • If want to present point as Right Eye L1 being behind screen: (P2) use L2, R2 Near side of Far side of Stereo window image plane image plane • Points L1 & R1 are homologous (same point in Adapted from image space) [Slater, Steed, Chrysanthou]
Setting up a stereo view: Parallax • IF: R-L > 0 : (P2) Positive horizontal parallax: -1 1 Left Eye Points will be virtual points R1 behind the stereo window. • IF: R-L < 0 : (P1) L2 P1 P2 R2 Negative horizontal parallax: Points will be virtual points in Right Eye L1 front of the stereo window. In front of Behind stereo stereo window window Adapted from [Slater, Steed, Chrysanthou]
Viewing Stereo pairs – what does it mean for your eyes? Crossed • Parallel setup: set up – Right eye sees Right image, Left Eye R1 – Left eye sees left image – Requires focus beyond the L2 Parallel P1 images set up R2 Right Eye • Crossed setup: L1 – Right eye sees left image In front of Behind stereo – Left eye sees right image stereo window window – Requires crossing eyes • Viewing the opposite way around will reverse the sense of depth. Adapted from [Slater, Steed, Chrysanthou]
Presenting 3D images: Ideals • Congruence L &R images should be same (except as caused by the horizontal parallax) Especially colour & brightness same for homologous points • Vertical parallax = Zero If>0, uncomfortable to fuse images • Parallax (view separation) trade off… Wide parallax: good depth, but too wide leads to discomfort. Parallax should be less <= IPD Closer the homologous points…less disparity between convergence & accommodation To provide maximum depth but lowest parallax, place principal objects so that ~½ parallax values are +ve, ½ -ve
CAVE-like Displays • Screens surround the user • Modelled as a series of cameras, two per wall • Each camera defined by corners of the wall and centre of eye
CAVE Projection X Y (out of screen) Z Straight forward to show that: Adapted from [Cruz-Neira et al]
Cruz-Neira et al.’s Discussion • Advantages of a CAVE over HMD or semi- immersive VR systems – Wide field of view – Less rotational instability – See yourself – Higher quality images – Less optical distortion
Cruz-Neira et al.’s Discussion • Disadvantages – Expensive and complex to configure • Need to align several projectors • Very high refresh rate needed (>100Hz) – User occludes the screen with their own body – Other users can occlude screen – Floor shadows – Typically 4 walls – Only one perspective-correct user at once
Summary • Current display technology is limited – Colour, brightness, frame-rate, contrast • However 3D computer graphics is very successful • Brain uses a variety of cues – Physiological – Psychological • Properly configured immersive systems portray robust illusions of 3D objects
VR Systems UK
Next Week! • Lab session on Tuesday from 2-4pm • Lecture (given by Mel Slater) on Friday, time to be confirmed.
Recommend
More recommend
