Virtual Rear Projection Technology Evaluation Jay Summet summetjcc

Virtual Rear Projection: Technology & Evaluation Jay Summet summetj@cc. gatech. edu

Introduction Jay Summet - Ph. D student, Georgia Institute of Technology Co-Advised: Gregory Abowd (HCI / Ubicomp) & Jim Rehg (Computer Vision) Other work: – Tracking and Projecting on handheld displays (Pervasive 2005, UIST 2005), – Detecting camera phones and blinding them (Ubicomp 2005) 2

Virtual Rear Projection Using multiple redundant front projectors to emulate the experience of a rear projected surface. • Introduction • Motivation for VRP • Initial Technology Development • User Evaluation • More Technology Development • Future Work 3

Rear Projection • No shadows! • But extra costs. . . – Display Material – Installation – Space cost (77$ sq. ft. ) – Immobile 4

Larger Board = Higher Cost 5

Front Projection • Inexpensive: – Display Screen – Installation – Mobility • Effective use of space. • But shadows & blinding light are annoying! 6

Shadows 7

Blinding Light 8

Warped Front Projection (WFP) • Moves shadow away from directly in front of the user. • Commercial products using WFP: – NEC WT 600 – 3 M Idea. Board 9

WFP Measurements 3379 166 10

Passive VRP (PVRP) • Overlapped projectors fill in shadows. • Calibration via camera or manually. • Projective transforms done on graphics card. 11

Passive VRP Measurements 2509 167 12

Movie (part 1) Demo Movie of WFP/PVRP 13

Benefits of Redundant Illumination One Projector (WFP) Two Projectors (PVRP) 14

Research Questions • Are shadows / blinding light a problem? – Very little research with interactive surfaces performed using front projection. – But no real research into the effects of shadows on users of interactive surfaces. • Is Passive VRP “good enough”? 15

Projection Technologies Studied Front Projection Virtual Rear Projection Warped Front Projection Rear Projection 16

Participants 17 Participants – Undergraduate students – Mean age: 21. 3 Std. Dev 1. 77 – 9 males, 8 females – Exclusively right handed – Normal or corrected-to-normal vision 17

Task Box Task – 8 starting positions – Target in Center Dependent Variables – Acquire time – Total Time – Number of occluded boxes 18

Results (1/3) Subjective: – Users found projected light annoying – Users had clear technology preferences: FP, WFP < VRP < RP 19

Results (2/3) Quantitative: – Box Acquire Time Slower: FP < WFP, VRP < RP – Less Boxes Occluded FP – 178 WFP – 66 VRP – 4 RP – 0 20

Results (3/3) Behavioral: – Users adopted coping behaviors to deal with shadows in the FP and WFP conditions – Not present in the VRP and RP conditions • Edge of Screen – 7 • Near Center – 7 • Move on Occlusion – 3 • Dead Reckoning - 1 21

Movie Participant Video Figure 22

Edge of Screen (7 participants) 23

Near Center (7 participants) Participants would stand in the center and. . . –. . . either be short enough so that they would not occlude boxes. (3 participants) –. . . or they would sway their bodies to find occluded boxes. (4 participants) 24

Move on Occlusion (3 participants) These participants would move whenever they occluded a box, and stay there until they occluded another. 25

Findings (CHI 05) • Users prefer Rear Projected and Passive Virtual Rear Projected displays over the others. • RP and passive VRP eliminated coping behaviors seen in FP and WFP. • Users find projected light to be annoying. • Passive VRP casts light on users. 26

Projected light is a larger problem as you add more projectors. 27

Technology Development Shadow Elimination – CVPR '01 – R. Sukthankar, T. -J. Cham, G. Sukthankar – U. Kentucky – C. Jaynes, Visualization 2001 28

Shadow Elimination Measurements 1052 221 29

Technology Development Blinding Light Suppression – CVPR '03 – Tat Jen. Cham, Jim Rehg, Rahul Sukthankar, Gita Sukthankar 30

SE + BLS Measurements 1165 34 31

Interesting, but useless • Required an unoccluded view of the screen, too slow. 32

Technology Development Switching – PROCAMS '03 – Ramsaroop Sommani GPU Enhancements – PROCAMS '05 – Matt Flagg 33

Active Virtual Rear Projection • Detects occluders, turns off pixels they are occluding, and fills in those pixels with alternate projectors 34

Active VRP Measurements 1466 12 35

Movie (part 2) Active VRP 36

Future Work User evaluation of Active VRP – Controlled laboratory study (80 participants) – Exploratory Research • Aero. Space Engineering Design Lab • “Home-Office” in Aware Home 37

More information: summetj@cc. gatech. edu http: //www. cc. gatech. edu/cpl/vrp http: //www. cc. gatech. edu/cpl/proc ams 38

Thank you! The End 39

Table of Relative Performance 40