Improving drag-and-drop on wall-size displays

1
Improving drag-and-drop on wall-size displays

• Maxime Collomb, LIRMM
• Patrick Baudisch, Microsoft Research
• Mountaz Hascoët, LIRMM
• Brian Lee, Stanford Computer Graphics

2
Push-and-pop not quite a demo
Download our demo from www.lirmm.fr/vag/dragging
3
Warning

• Lots of X-and-Y names ahead

4
Contents

• Pointing on wall-size displays
• Survey of existing techniques
• Picking push-and-throw drag-and-pop
• The best of both worlds push-and-pop
• User Studies
• Conclusion

5
Wall-size displays

• Augmented surfaces are more and more widely
  available
• Smartboard
• DynaWall
• iWall

6
Problems with pentouch

• user may not be able to reach content
• Target may be too high
• Bezels not touch sensitive
• Fatigue

7

• Related work(RSVP version)

8
Pick-and-drop

• Pick-and-drop Rekimoto, 97
• Take-and-put Geißler, 98

9
Hyperdragging Rekimoto, CHI 99
10
Frisbee Khan, UIST 04
11
Laser pointers Olsen 01, Myers 02
12
TractorBeam Parker, in 30min
13
Push-and-throw Hascoët, HCI 03

• Pantograph principle(go go technique Poupyrev,
  UIST 96)

14
RADAR Nacenta, CHI 05

• Pantographs (lots of them)

15
Drag-and-popBaudisch, INTERACT 03
16
Vacuum Bezerianos, CHI 05
17

• The question

18

• What is better
• Bring user to content (push-and-throw)
• Bring content to user (drag-and-pop)
• or just good old direct manipulation(pick-and-dr
  op)?

19
Fixing drag-and-pop

• Use asymmetric target sector
• Make sure proxies appear in current display unit

20
Fixing push-and-throw

• Add non-linear acceleration
• Add semantic pointingslow down pointer over
  target Blanch CHI 04

21

• But wait maybe push-and-throw and drag-and-pop
  are not mutually exclusive

22
push-and-pop

• From drag-and-pop
• Bringing targets to the user
• ?Avoid imprecision of push-and-throw
• From push-and-throw
• Using the object centered take-off area
• ?Avoid target set selection error of drag-and-pop

23
push-and-pop
24

• User studies

25
? iWall(exp. n2)6 participants
? Double display(exp. n1)12 participants
26
Interfaces
Technique Approach Need to reorient
Drag-and-drop Never
Pick-and-drop Never
Push-and-throw To target Constantly
Accelerated PT To target Constantly
Drag-and-pop To pointer Once
Push-and-pop To pointer Once, later never
27
Task

• Dragging icons into target iconon a simulated
  desktop
• Independent variables
• 6 interfaces
• 12 distances
• repeated 3 times(4 times on the 2nd exp.)
• 216 / 288 trials for each participant
• Dependent variables
• Movement time
• Error rate
• Subjective satisfaction

28
First study results
Mean task completion time for each technique
depending of the ID of the task
29
First study results
Error rates (in percent)
30
First study results
Subjective preferences (higher is better)
31

• Repeated same study on 3-unit display wall

32
Second study results
Mean task completion time for each technique
depending of the ID of the task
33
Second study results
Error rates (in percent)
34
Second study results
Subjective preferences (higher is better)
35
Discussion

• Drag-and-drop performs well on single display
• Among pointer-to-target techniques, acc.
  push-and-throw design improvements were
  beneficial
• Push-and-pop outperformed all other techniques

36
Conclusion

• Techniques that bring content to the user
  performed better
• We think this is because they minimize the need
  for users to reorient themselves

37
Future work

• Extends Push-and-pop to widgets
• Optimize design of push-and-pop

38
Future work

• Extends Push-and-pop to widgets

39
Thank you!

• Thanks to Mountaz Brian
• Download our demo from
• www.lirmm.fr/vag/dragging

40
 In case of  slides

1. The  many same-icon targets  problem
2. Pnp with rubber bands
3. Pnp with one rubber band (current selected
   target)
4. Experiment desktop layout choice
5. Experiment rooms arrangement
6. Extending push-and-pop to widgets

41
The  many same-icon targets  problem
42
Push-and-pop with rubber bands
43
Push-and-pop with one rubber band (current
selected target)
44
Experiment desktop layout choice (1)

• Test
• 4 blocks of 10 trials
• 2 interaction techniques
• drag-and-drop
• push-and-pop
• 2 desktop layouts
• Same sources and targets (different positions for
  each trial).
• Different source and targets.
• A total of 160 trials
• Single display and mouse were used

45
Marking menus Kurtenbach, CHI 94
46
Experiment desktop layout choice (2)

• Result
• Differences wasnt significant between the 2
  desktop layouts
• P 0.706 for drag-and-drop
• P 0.566 for push-and-pop

47
Future work Tablecloth