A Comparison of Direct and Indirect Multi-touch Input for Large Surfaces

Dominik Schmidt, Florian Block, Hans GellersenLecture Notes in Computer Science 2009 · DOI · OpenAlex

Introduction

Multi-touch interaction is traditionally synonymous with direct manipulation, where input and display are co-located. However, direct touch on large surfaces often leads to screen occlusion and physical strain. This study, presented at INTERACT 2009, investigates "indirect" multi-touch—separating the input surface from the display—to quantify the performance costs and behavioral shifts associated with this spatial offset.

Methodology

The researchers conducted a controlled experiment with 22 participants to compare two interaction conditions:

  1. Direct: Users interacted directly with a horizontal tabletop display.
  2. Indirect: Users touched the horizontal surface while viewing a separate vertical plasma screen.

To facilitate indirect interaction, a ceiling-mounted camera tracked hand movements, projecting real-time "hand contours" and touch points onto the vertical screen to provide visual feedback.

The indirect interaction setup showing hand contours projected onto a vertical display.

Participants performed a symmetric bimanual task involving pointing (selecting handles on a square target) and dragging/resizing (matching the target to a "dock"). The study utilized a repeated-measures design, varying target size ($w$) and travel distance ($d$) to measure performance across 5,940 interaction trials.

Diagram of the experimental task, illustrating the target, dock, and home area parameters.

Discussion of Findings

The study revealed several critical insights into how spatial separation affects user behavior:

  • The "Hovering" Penalty: Indirect mode lacks the proprioceptive ease of direct touch. Users often performed "sequential" movements—aligning hands in the air before touching the surface—rather than the fluid, ballistic motions seen in direct interaction.
  • Task Asymmetry: The performance gap was most pronounced during pointing, which took 200% longer in the indirect condition. Conversely, dragging was only 23% slower, suggesting that continuous visual feedback helps users compensate for spatial offsets once an object is acquired.
  • Cognitive and Physical Load: Indirect interaction was found to be more mentally taxing, with users reporting higher frustration. Physically, users adopted a tense "claw" posture to ensure tracking accuracy, leading to increased fatigue.

Key Results

  • Performance: Direct interaction was significantly faster (Mean = 4.3s) than the indirect condition (Mean = 6.7s). Pointing errors were three times more frequent in the indirect condition.
  • Sensitivity: Indirect touch was more sensitive to task difficulty; smaller targets and longer distances resulted in disproportionately higher performance penalties compared to direct touch.
  • User Preference: While 19 out of 22 participants preferred direct touch, proponents of the indirect setup highlighted improved visibility (no hand occlusion) and a more ergonomic "head-up" posture.

Conclusion and Future Directions

While direct multi-touch remains the gold standard for speed and accuracy, indirect interaction is a viable alternative for environments where occlusion or ergonomics are primary concerns. The authors suggest that future work should focus on gesture sets specifically designed for indirect surfaces—such as "sliding" into targets rather than tapping—and improving hardware to allow users to rest their hands without triggering accidental inputs, which would mitigate the fatigue of hovering.

Summary updated

BibTeX
@inproceedings{schmidt2009comparison,
  author = {Dominik Schmidt and Florian Block and Hans Gellersen},
  title = {{A Comparison of Direct and Indirect Multi-touch Input for Large Surfaces}},
  booktitle = {Lecture Notes in Computer Science},
  year = {2009},
  doi = {10.1007/978-3-642-03655-2\_65},
}
Figure