user issues in 3d tv & cinema

49
User Issues in 3D TV & Cinema Martin S. Banks Vision Science Program UC Berkeley

Upload: sveta

Post on 23-Feb-2016

44 views

Category:

Documents


0 download

DESCRIPTION

User Issues in 3D TV & Cinema. Martin S. Banks Vision Science Program UC Berkeley. Issues in 3D TV & Cinema. Technical Issues Developing content Sufficient resolution over time: temporal aliasing Sufficient separation between two eyes’ images: “ghosting” User Issues - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: User Issues in 3D TV & Cinema

User Issues in 3D TV & Cinema

Martin S. BanksVision Science Program

UC Berkeley

Page 2: User Issues in 3D TV & Cinema

Issues in 3D TV & Cinema

Technical Issues• Developing content• Sufficient resolution over time: temporal aliasing• Sufficient separation between two eyes’ images: “ghosting”

User Issues• Perceptual distortions due to incorrect viewing position• Flicker & motion judder due to temporal sampling• Maintaining depth across scene cuts• Window violations• Residual ghosting• Visual discomfort due to vergence-accommodation conflict• Appropriate blur relative to other depth signals• Conflict between visually-induced motion & vestibular signals

Page 3: User Issues in 3D TV & Cinema

Technical Issues• Developing content• Sufficient resolution over time: temporal aliasing• Sufficient separation between two eyes’ images: “ghosting”

User Issues• Perceptual distortions due to incorrect viewing position• Flicker & motion judder due to temporal sampling• Maintaining depth across scene cuts• Window violations• Residual ghosting• Visual discomfort due to vergence-accommodation conflict• Appropriate blur relative to other depth signals• Conflict between visually-induced motion & vestibular signals

Issues in 3D TV & Cinema

Page 4: User Issues in 3D TV & Cinema

Technical Issues• Developing content• Sufficient resolution over time: temporal aliasing• Sufficient separation between two eyes’ images: “ghosting”

User Issues• Perceptual distortions due to incorrect viewing position• Flicker & motion judder due to temporal sampling• Maintaining depth across scene cuts• Window violations• Residual ghosting• Visual discomfort due to vergence-accommodation conflict• Appropriate blur relative to other depth signals• Conflict between visually-induced motion & vestibular signals

Issues in 3D TV & Cinema

Page 5: User Issues in 3D TV & Cinema

Foca

l dis

tanc

e

Verg

ence

dis

tanc

eVergence & Accommodation: Natural Viewing

Page 6: User Issues in 3D TV & Cinema

3 6

1.5

3

4.5

6

Vergence Distance (diopters)00

Foca

l Dis

tanc

e (d

iopt

ers)

Foca

l dis

tanc

e

Verg

ence

dis

tanc

e

1.5 4.5

Vergence & Accommodation: Natural Viewing

Page 7: User Issues in 3D TV & Cinema

3 6

1.5

3

4.5

6

Vergence Distance (diopters)00

zone of clear singlebinocular vision

Foca

l Dis

tanc

e (d

iopt

ers)

Foca

l dis

tanc

e

Verg

ence

dis

tanc

e

1.5 4.5

Vergence & Accommodation: Natural Viewing

Page 8: User Issues in 3D TV & Cinema

3 6

1.5

3

4.5

6

Vergence Distance (diopters)00

Percival's zoneof comfort

zone of clear singlebinocular vision

Foca

l Dis

tanc

e (d

iopt

ers)

Foca

l dis

tanc

e

Verg

ence

dis

tanc

e

1.5 4.5

Vergence & Accommodation: Natural Viewing

Page 9: User Issues in 3D TV & Cinema

Foca

l dis

tanc

e

Verg

ence

dis

tanc

eVergence & Accommodation: Stereo Display

Page 10: User Issues in 3D TV & Cinema

3 6

1.5

3

4.5

6

Vergence Distance (diopters)00

Foca

l Dis

tanc

e (d

iopt

ers)

1.5 4.5

Foca

l dis

tanc

e

Verg

ence

dis

tanc

eVergence & Accommodation: Stereo Display

Percival's zoneof comfort

zone of clear singlebinocular vision

Page 11: User Issues in 3D TV & Cinema

Displays with Nearly Correct Focus Cues

Two multi-focal displays we’ve developed:1.Fixed-viewpoint, volumetric display with mirror system & 3 focal

planes (Akeley, Watt, Girshick, & Banks, SIGGRAPH, 2004).

2.Fixed-viewpoint, volumetric display with switchable lens & 4 focal planes (Love, Hoffman, Kirby, Hands, Gao, & Banks, Optics Express, 2009)

Page 12: User Issues in 3D TV & Cinema

Multi-focal Display

Akeley, Watt, Girshick & Banks (2004), SIGGRAPH.

Page 13: User Issues in 3D TV & Cinema

Akeley, Watt, Girshick & Banks (2004), SIGGRAPH.

Multi-focal Display

Page 14: User Issues in 3D TV & Cinema

Akeley, Watt, Girshick & Banks (2004), SIGGRAPH.

Multi-focal Display

Page 15: User Issues in 3D TV & Cinema

Depth-weighted Blending

• Depth-weighted blending along lines of sight

• Weights dependent on dioptric distances to planes

Akeley, Watt, Girshick, & Banks (2004), SIGGRAPH.

Page 16: User Issues in 3D TV & Cinema

Do V-A Conflicts Cause Fatigue/Discomfort?

Page 17: User Issues in 3D TV & Cinema

• 600-ms stimulus at near or far vergence-specified distance• Appeared at each focal distance

Hoffman, Girshick, Akeley, & Banks (2008), Journal of Vision

Do V-A Conflicts Cause Fatigue/Discomfort?

Page 18: User Issues in 3D TV & Cinema

** ** ** **

cues-inconsistentcues-consistent

Sev

erity

of S

ympt

om

1

3

5

7

9

How tired are your eyes?

How clear is your vision?

How tired or sore are your

neck & back?

How do your eyes feel?

How does your head feel?

** = p < 0.01 (Wilcoxen test)

Hoffman, Girshick, Akeley, & Banks (2008), Journal of Vision

Do V-A Conflicts Cause Fatigue/Discomfort?

Page 19: User Issues in 3D TV & Cinema

*******Which session was

more fatiguing?

Which session irritated

your eyes more?

Which session gave you

more headache?

Which session did

you prefer?

cues-consistent much worse than

inconsistent

cues-inconsistent much worse than

consistent

no difference

** = p < 0.01 (Wilcoxen test)

Hoffman, Girshick, Akeley, & Banks (2008), Journal of Vision

Do V-A Conflicts Cause Fatigue/Discomfort?

Page 20: User Issues in 3D TV & Cinema

Discomfort & 3D Cinema

Page 21: User Issues in 3D TV & Cinema

Discomfort & 3D Cinema

Page 22: User Issues in 3D TV & Cinema

Discomfort & 3D Cinema

Page 23: User Issues in 3D TV & Cinema

Discomfort & 3D Cinema

Page 24: User Issues in 3D TV & Cinema

Discomfort & 3D Cinema & TV

Page 25: User Issues in 3D TV & Cinema

Technical Issues• Developing content• Sufficient resolution over time: temporal aliasing• Sufficient separation between two eyes’ images: “ghosting”

User Issues• Perceptual distortions due to incorrect viewing position• Flicker & motion judder due to temporal sampling• Maintaining depth across scene cuts• Window violations• Residual ghosting• Visual discomfort due to vergence-accommodation conflict• Appropriate blur relative to other depth signals• Conflict between visually-induced motion & vestibular signals

Issues in 3D TV & Cinema

Page 26: User Issues in 3D TV & Cinema

Almost never view pictures from correct position.

Retinal image thus specifies different scene than depicted.

Do people compensate, and if so, how?

Viewing Pictures

Page 27: User Issues in 3D TV & Cinema

Stimuli

Vishwanath, Girshick, & Banks (2005), Nature Neuroscience.

Page 28: User Issues in 3D TV & Cinema

Stimulus: simulated 3D ovoid with variable aspect ratio.Task: adjust ovoid until appears spherical.

Vary monitor slant Sm to assess compensation for oblique viewing positions.

Spatial calibration procedure.If compensate, will set ovoid to sphere on screen (ellipse on retina).

Observation Point

Sm

CRT

Experimental Task

Vishwanath, Girshick, & Banks (2005), Nature Neuroscience.

Page 29: User Issues in 3D TV & Cinema

Center of Projection

Observation Point

No compensation: set ovoid to make image on retina circular:

retinal coordinates

screen coordinates

Predictions

Page 30: User Issues in 3D TV & Cinema

Center of Projection

Observation Point

Compensation: Set ovoid to make image on screen circular:

Predictions

retinal coordinates

screen coordinates

Page 31: User Issues in 3D TV & Cinema

Asp

ect R

atio

(scr

een

coor

ds)

1

1.2

1.4

-40 -20 0 20 40

invariance predictions

Viewing Angle Sm (deg)

Sm

Predictions

Page 32: User Issues in 3D TV & Cinema

Predictions

1

1.2

1.4

-40 -20 0 20 40

invariance predictionsretinal predictions

Asp

ect R

atio

(scr

een

coor

ds)

Viewing Angle Sm (deg)

Sm

Page 33: User Issues in 3D TV & Cinema

Results

1

1.2

1.4

-40 -20 0 20 40

monoc-aperture

invariance predictsretinal predicts

Asp

ect R

atio

(scr

een

coor

ds)

Viewing Angle (deg)

JLL

Vishwanath, Girshick, & Banks (2005), Nature Neuroscience.

Page 34: User Issues in 3D TV & Cinema

Results

1

1.2

1.4

-40 -20 0 20 40

monoc-aperture

binoc-no aperture invariance predictsretinal predicts

JLL

Results

Asp

ect R

atio

(scr

een

coor

ds)

Viewing Angle (deg)Vishwanath, Girshick, & Banks (2005), Nature Neuroscience.

Page 35: User Issues in 3D TV & Cinema

Compensation for Incorrect Viewing Position

• Pictures not useful unless percepts are robust to changes in viewing position.

• People compensate for oblique viewing position when viewing 2d pictures.

• Two theories of compensation: pictorial & surface. Data clearly favor surface compensation.

• Two versions of surface method: global & local. Data clearly favor local slant.

Page 36: User Issues in 3D TV & Cinema

2D Pictures vs 3D Pictures

• Two eyes presented same image

• Binocular disparities specify orientation & distance of picture surface; hence useful for compensation

2D

Page 37: User Issues in 3D TV & Cinema

• Two eyes presented different images

• Binocular disparities specify orientation & distance of picture surface and layout of picture contents; hence not useful for compensation

• Two eyes presented same image

• Binocular disparities specify orientation & distance of picture surface; hence

useful for compensation

3D

• Two eyes presented same image

• Binocular disparities specify orientation & distance of picture surface; hence useful for compensation

2D

2D Pictures vs 3D Pictures

Page 38: User Issues in 3D TV & Cinema

Stereo (3D) Pictures

• For most applications, viewers will not be at correct position.

• Retinal disparities thus specify a different layout than depicted.

• Do people compensate?• Is correct seating position for a

3D movie more important than for 2D movie?

Page 39: User Issues in 3D TV & Cinema

Stereo Picture Geometry

display surface

stereo projectors

Page 40: User Issues in 3D TV & Cinema

display surface

stereo projectors

depicted hinge

Stereo Picture Geometry

Page 41: User Issues in 3D TV & Cinema

display surface

stereo projectors

depicted hinge

Stereo Picture Geometry

Page 42: User Issues in 3D TV & Cinema

display surface

stereo projectors

depicted hinge

disparity-specified hinge

Stereo Picture Geometry

Page 43: User Issues in 3D TV & Cinema

perceiveddihedralangle?

display surface

stereo projectors

depicted hinge

disparity-specified hinge

Stereo Picture Geometry

Page 44: User Issues in 3D TV & Cinema

Predictions

0 25 450

30

60

90

120

Viewing Angle (deg)

35°

17.5°

-17.5°

-35°

Hin

ge S

ettin

g (d

eg) Invariance: Hinge settings

are 90° for all viewing angles and base slants

Retinal disparity: Hinge settings vary significantly with viewing angle & base slant

Page 45: User Issues in 3D TV & Cinema

Viewing Angle (deg)

0 25 45

30

60

90

120

Hin

ge S

ettin

g (d

eg)

Results

non-stereo pictures

Page 46: User Issues in 3D TV & Cinema

Viewing Angle (deg)

0 25 45

30

60

90

120

stereo pictures

0 25 45

30

60

90

120

Hin

ge S

ettin

g (d

eg)

Results

non-stereo pictures

Page 47: User Issues in 3D TV & Cinema

Viewing Angle (deg)

0 25 45

30

60

90

120

0 25 45

30

60

90

120

Hin

ge S

ettin

g (d

eg)

non-stereo pictures stereo pictures

Results

Page 48: User Issues in 3D TV & Cinema

Summary

• User issues in 3D cinema & TV• Vergence-accommodation conflicts cause visual fatigue &

discomfortCan be handled by attending to viewer’s distance from screen & range of disparities presented relative to screen

• Perceptual distortions due to incorrect viewing positionCompensation is good with non-stereo pictures

Compensation is significantly poorer with stereo pictures suggesting that viewer position could be more important

Page 49: User Issues in 3D TV & Cinema

Acknowledgments

• Kurt Akeley (Microsoft)• Simon Watt (Univ. of Wales, Bangor)• Ahna Girshick (NYU)• David Hoffman (UC Berkeley)• Robin Held (UC Berkeley)• Funding from NIH, NSF, & Sharp Labs