Visualizing Off-Screen Locations on Small Mobile Displays Sean Gustafson master of science thesis defense Dec 15 / 2008.

Visualizing Off-Screen Locations onSmall Mobile Displays

Sean Gustafson

master of sciencethesis defense

Dec 15 / 2008

2

off-screen viz

overview arrows halo

3

atedworkrel

4

overviews

[Plaisant 95 and many others]

5

fisheye views

fisheye views

[Carpendale 01]

6

simple arrows

[Tecmo Bowl 87]

7

scaled and stretched arrows

[Burigat 06]

8 [Gustafson 07]

edgeradar

9 [Baudisch 03]

halo

10 [Baudisch 03]

halo

11

the problem: clutter

12

the problem: clutter

13

corner

(also not great: corners)

side

14

cut lines color

ovalsclustering

Irani et al.Baudisch & Rosenholtz

initial approaches

separatevisually

15

ewedgeth

16

leg

legtarget

basebasebasebasewedge

17

halo wedge

18

3 degrees of freedom

rotationbase lengthintrusion

19

avoiding overlap

rotation

base length

intrusion

20

outalgorithmlay

21

avoidance algorithm

22

avoidance algorithm

23

avoidance algorithm

24

avoidance algorithm

25

avoidance algorithm

26

leaves 2 degrees

intrusion ~ distance

base-width ~ distance

27

serstudyu

28

two interface conditions

halo and wedge

29

halo

task 1: locate

wedge

same as original halo paper

30

halo

task 2: closest

wedge

same as original halo paper

31

halo

task 3: avoid

wedge

same as original halo paper

32

IV: density

sparse dense

33

IV: cornerside

corner

34

36 participants

35

1.wedge is more accurate

2.larger improvement in dense condition

3.larger improvement in corners

(no hypothesis about task time)

hypotheses

36

results

37

participants were significantly more accurate when using the wedge

38

error in pixels

locate task

Err

or

(pix

els) 60

40

20

0

SparseDense

Side

Wedge

Halo80

39

error in pixels

locate task

Err

or

(pix

els) 60

40

20

0

SparseDense

Side

Wedge

Halo80

Corner

SparseDense

40

other results

no interesting results from the other tasks or from completion time.

41

other resultsWedge Halo No

Preference

Locate 25 10 1

Avoid 21 13 2

Closest 18 14 4

preferences

42

1.wedge is more accurate

2.larger improvement in dense condition

3.larger improvement in corners

hypotheses

43

elingpointingmod

44

45

beam

46

beam

47

beam

beam

48

orbitalbeam

beam

49 intrusion

50 intrusion

51 aperture

52 aperture

53 rotation

54 rotation

55

experiment to gather accuracy data

data collection

56

experiment to gather accuracy data

data collection

distance

legLength1

legLength2 base

Leng

th

57

results50

100

200

300

Depth Error

Dis

tanc

e

-100 0 100 200 300-60

60

0

-60

60

0

-60

60

0

-60

60

0

Bre

ad

th E

rro

r

58

results

Depth Error

-100 0 100 200 300

60

0

-60bias

length

width

59

results - bias

60

results - length

61

minimizing errors

• choose smallest base and leg length that still maintains small orbital and low bias

• base length of at least 50px, 100px for longer distances

• a large leg length (above 8px) does not result in smaller orbital

• avoid heavily rotating wedges

62

conclusion

63

1. summary of existing techniques and criteria for evaluating them

2. introduction of Wedge

3. introduction of the concept of degrees of freedom and dynamic layout

4. evaluation of Wedge

5. preliminary model

contributions

64

thanks

65

66

IV: density2

3 4

1sparse

dense

67

68

correcting errors

69

correcting errors

70

71

fiedpointinguniwedge=

72

intrusion 0

citylights

73

aperture 0

stretched arrows

74

?halo

75

round bases

distance information in corner distance information along arc

76

halo

aperture 360

77

78 locate task

SparseDense

Err

or

(pix

els

) 60

40

20

0SparseDense

SideCorner

Wedge

Halo80

SparseDense

Tim

e (

ms)

3000

2000

1000

SparseDense

SideCorner

Wedge

Halo

0

full results

79 avoid task

full results

SparseDense

Err

or

Ra

te %

50

40

30

20

10

0SparseDense

Wedge

Halo

SideCorner

Tim

e (

ms)

5000

4000

2000

0

SideCorner

Wedge

Halo

SparseDense SparseDense

3000

1000

80 closest task

full results

SparseDense

Err

or

Ra

te %

60

40

20

0SparseDense

WedgeHalo

SideCorner

SparseDenseT

ime

(m

s)

6000

4000

2000

0

SparseDense

SideCorner

WedgeHalo

81