building an autostereoscopic display

21
Building an Autostereoscopic Display CS448A – Digital Photography and Image- Based Rendering Billy Chen

Upload: deidra

Post on 09-Feb-2016

39 views

Category:

Documents


0 download

DESCRIPTION

Building an Autostereoscopic Display. CS448A – Digital Photography and Image-Based Rendering Billy Chen. Original Goals. dynamic, real-time display convenient 3D display for the home (3D desktops) autostereoscopic light field viewer. Display design choices. Physical Setup. Render. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Building an Autostereoscopic Display

Building an Autostereoscopic Display

CS448A – Digital Photography and Image-Based Rendering

Billy Chen

Page 2: Building an Autostereoscopic Display

Original Goals

• dynamic, real-time display

• convenient 3D display for the home (3D desktops)

• autostereoscopic light field viewer

Page 3: Building an Autostereoscopic Display

Display design choicesDisplay Type Resolution Advantage Disadvantage

CRT/LCD 72-115 dpi cheap, everywhere low resolution, challenging calibration

Projector ~150 dpi easier to control pixel depth, adjustable angular resolution

projector distortion

Big Bertha 211.66 dpi higher resolution expensive, need special hardware to drive it

Printer 300 dpi high resolution static images

Page 4: Building an Autostereoscopic Display

Physical Setup

Page 5: Building an Autostereoscopic Display

Overview of display process

Render ` Calibration

Page 6: Building an Autostereoscopic Display

The calibration problem

+

Page 7: Building an Autostereoscopic Display

Calibration

• affine transformation correction (mostly scale)• projective transformation correction

Page 8: Building an Autostereoscopic Display

Calibration solution 1

• OpenGL program displays a moiré pattern• can calibrate up to affine transformations• most effective for finding correct size

Page 9: Building an Autostereoscopic Display

Calibration solution 2p

h

p

Finding the homography without getting projector parameters

A

B

A’

B’

Page 10: Building an Autostereoscopic Display

xy1

cx’cy’c

Calibration solution 2M =

M p cp’Let Mi = i’th row of M

(1) M1p = cx’(2) M2p = cy’(3) M3p = c

y’ (M1p) - x’(M2p) = 0

M1p - x’(M3p) = 0

xy’ yy’ y’ -xx’ -yx’ -x’ 0 0 0 M11

M12

M13

M21

...

...

= 0

8x9 9x1A=Take SVD(A) and look at matrix

Page 11: Building an Autostereoscopic Display

Calibration solution 2

Page 12: Building an Autostereoscopic Display

Rendering

• sampling the light field

• computing lenslet distances

• cropping and compositing

Page 13: Building an Autostereoscopic Display

Rendering: Sampling a light field

Isaksen et al., Siggraph 2000

Page 14: Building an Autostereoscopic Display

Getting “floating” images

Halle, Kropp. SPIE ‘97

Page 15: Building an Autostereoscopic Display

Sampling a light field

Page 16: Building an Autostereoscopic Display

Rendering: computing the FOV

Page 17: Building an Autostereoscopic Display

Rendering: compositing and cropping images

subsample crop

composite

Page 18: Building an Autostereoscopic Display

Implementation Details• Fresnel hex array #300; 0.12 in. focal length, 0.12 in. thickness, .09 in. diameter• default size for a lenslet image: 26x31 pixels (for 300 dpi displays)• calibrate scale is .49 (sanity check: 300 dpi / 150 dpi)• OpenGL unit == 1 pixel (300 dpi)• SEE WEBPAGE!

Page 19: Building an Autostereoscopic Display

Results compared to original goals

• real-time display is hard, must handle the bandwidth

• spatial resolution too small for 3D desktops

• light fields have problems with much depth complexity, but NEED depth for effective autostereoscopic displays

Page 20: Building an Autostereoscopic Display

Future Work

• reflective display

• auto-calibration

• hardware accelerated light field sampling

• overloading pixels per direction: perspective views, displacing display pixels from focal plane

• use a light field of captured data

Page 21: Building an Autostereoscopic Display

Acknowledgements

• calibration: Vaibhav Vaish• light field generator: Georg Petschnigg• hardware accelerated approach: Ren Ng• bootstrap: Sean Anderson