PaperlayerView exact revision on arXiv

Source-linked AI summary

UniScale: Exploring Unimanual Gesture Mapping Strategies for Gaze+Pinch-based Scaling Interaction

Kyoungwhan Mheen, Jinwook Kim, Sang Ho Yoon

arXiv:2609.00500v1cs.HC

TL;DR

XR scaling remains constrained by bimanual interaction that occupies both hands and can increase fatigue. This paper introduces and evaluates five unimanual Gaze+Pinch mapping strategies against a bimanual baseline under clutching and clutching-free conditions. The results show that clutching preference depends on the mapping, with isomorphic techniques favoring discrete clutching and rate-based techniques favoring clutching-free control.

  • Problem

    XR spatial manipulation lacks scaling functions that integrate seamlessly with Gaze+Pinch while preserving hand availability and avoiding bimanual fatigue.

  • Method

    UniScale proposes five indirect unimanual scaling techniques based on depth, angle, micro-gestures, and finger-distance mappings, evaluated against a bimanual baseline in a 3D resizing task.

  • Results

    Isomorphic mappings benefited from discrete clutching, whereas rate-based mappings excelled in continuous clutching-free modes; users accepted minor precision compromises for significantly reduced physical fatigue.

  • Takeaways & Limitations

    UniScale indicates that clutching should be matched to the scaling mapping, while unimanual control can free the dominant hand for more efficient spatial multitasking.

  • Takeaways & Limitations

    The study evaluated scaling and translation sequentially, restricted magnification to 3×, and used a simplified scene with one interactable object at a time.

Abstract

from arXiv · show

Object scaling serves as a fundamental spatial manipulation that enables complex and productive tasks in XR environments. This paper investigates unimanual scaling techniques for XR using gaze and hand interactions. We propose UniScale, a set of unimanual alternatives to the standard bimanual pinch, allowing users to scale objects while preserving hand availability for concurrent spatial manipulations. We design five distinct mapping strategies based on physical metaphors, exploring unimanual control that varies depth, angle, micro-gestures, and finger-distance input. We then compare these techniques against a standard bimanual baseline, in which users adjust the inter-hand distance via a bimanual pinch gesture. In a user study, we evaluate their effectiveness in a 3D object scaling task under both clutching and clutching-free conditions. The results indicate that while bimanual scaling relies on clutching for stable control, unimanual techniques excel in clutching-free conditions, significantly reducing physical hand movement. From the results, we derive valuable design implications for developing efficient 3D multimodal interactions in XR.

1 INTRODUCTION

UniScale addresses the limited support for spatial manipulation in Gaze+Pinch XR by exploring unimanual scaling strategies that preserve hand availability. A study compares five techniques with a conventional bimanual baseline across clutching conditions.

  • Gaze+Pinch is widely adopted for intuitive, ergonomic target acquisition, but research has focused mainly on foundational selection mechanics.
  • Scaling is essential for productive XR spatial tasks, yet Gaze+Pinch lacks seamlessly integrated rotation, scaling, and translation functions.
  • Bimanual symmetric pinch occupies both hands, interrupts transitions to translation or rotation, and increases fatigue through repetitive clutching.
  • UniScale introduces five unimanual techniques using indirect Gaze+Pinch interaction and distinct physical metaphors for scaling.
  • The study compares UniScale with the commercially adopted bimanual baseline in a 3D resizing task using discrete clutching and continuous clutching-free conditions.
  • Isomorphic mappings benefit from discrete clutching, whereas rate-based mappings perform better in continuous clutching-free modes; suitable contexts are characterized from performance and feedback.

2 RELATED WORK

Related work establishes gaze and pinch as complementary tools for indirect XR interaction, while bimanual and unimanual manipulation present different ergonomic and expressive trade-offs.

  • Indirect manipulation decouples input from feedback, enabling interaction with distant objects without requiring locomotion.
  • Ray casting can lose precision at distance, while gaze improves rapid pointing but cannot by itself resolve depth ambiguity.
  • Gaze-hand systems have refined distant manipulation, and pinch has emerged as a preferred robust input gesture across these approaches.
  • Bimanual manipulation can be symmetric or asymmetric, whereas unimanual interaction offers a different balance between expressiveness and ergonomics.
  • Sustained bimanual mid-air interaction increases cumulative muscle load, while unimanual techniques can substantially reduce the gorilla-arm effect.

3 UNISCALE INTERACTION TECHNIQUE DESIGN

UniScale designs gaze-activated, unimanual scaling mappings across depth, angle, micro-gestures, and pinch span, then examines how clutching structures control and physical effort.

  • Interaction overview: UniScale uses gaze to indicate the target before mode activation, after which hand gestures govern scale manipulation.
  • Interaction overview: The design assigns scaling to the non-dominant hand so the dominant hand can remain available for spatial translation and other tasks.
  • Mapping strategies: uniDepth maps non-dominant-hand forward-backward motion to scale through an affordance-driven push-pull metaphor, but its reach demands may require repeated clutching.
  • Mapping strategies: uniDepth computes depth displacement as ∆z = zcurrent − zinitial, converts it to r = 1 − ∆z, and updates scale with Snew = Sinitial × r.
  • Mapping strategies: uniAngle maps vertical or angular hand movement to scale, keeping movement within arm reach but using a potentially unconventional metaphor.
  • Mapping strategies: uniMicro uses small non-dominant-thumb tap and swipe motions to reduce movement amplitude, while depending heavily on stable hand tracking.
  • Mapping strategies: uniSemi directly maps non-dominant index-thumb pinch distance to scale, whereas biSemi separates selection and scaling across hands using the same isomorphic mapping.
  • Clutching mechanisms: Clutching permits repositioning at physical boundaries but adds repositioning overhead; clutching-free scaling remains continuous while activation is maintained.

4 EVALUATION

The evaluation compares gaze-and-pinch scaling techniques in a 3D VR resizing-and-translation task, using six measures across clutching conditions. Participants completed randomized within-subject trials after training, with objective, workload, preference, and physical-movement measures collected.

  • Evaluation Setup: The baseline biDistance maps object size proportionally to the user’s inter-hand distance during a bimanual pinch.Selection begins with gaze and stable pinches in both hands; releasing either pinch terminates manipulation.
  • Task: The task required matching a blue control cube to one of four target sizes, then translating the matched object to a displayed white sphere.Targets were 0.333 m, 0.577 m, 1.732 m, and 3 m from an initial 1 m cube; completion required ±0.05 m accuracy for 500 ms.
  • Procedure: The study used a randomized within-subjects design, with at least five training sessions before testing each technique.Participants performed the scaling task under discrete clutching and continuous clutching-free conditions, followed by translation.
  • Measures: The evaluation measured completion, acquisition, fine-tuning, error, attempts, physical hand movement, usability, workload, satisfaction, rankings, and interview preferences.Physical hand movement was the cumulative Euclidean palm-center distance, summed across both palms for bimanual techniques.

5 RESULT

The results analyze six scaling techniques across clutching and clutching-free conditions using repeated-measures statistical tests. Task completion time differed by technique and showed a technique-by-clutching interaction.

  • 5 RESULT: 176 of 4,800 trials were excluded because of excessive final-size error or procedural anomalies.The analyzed design included scaling technique and clutching type as within-subject factors.
  • 5.1 Task Completion Time: Task completion time showed a significant technique effect, F(5,207) = 67.320, p < .001, η2p = .62.uniDepth was fastest at M = 2.568, followed by uniAngle at M = 2.606 and biDistance at M = 2.781.
  • 5.1 Task Completion Time: Task completion time showed a significant technique-by-clutching interaction, F(5,207) = 13.113, p < .001, η2p = .24.uniDepth, uniAngle, and uniMicro were faster clutching-free, whereas biDistance and biSemi were slower clutching-free; uniSemi did not differ significantly.

5.2 Initial Acquisition Time (Figure 5 (b))

Initial acquisition time differed across techniques and clutching conditions. uniAngle and uniDepth were faster than baseline, while clutching was faster overall and interacted significantly with technique.

  • 5.2 Initial Acquisition Time: Initial acquisition time showed a significant technique effect, F(5,207) = 34.2433, p < .001, η2p = .45.uniAngle and uniDepth were significantly faster than baseline, whereas uniSemi and biSemi were slower.
  • 5.2 Initial Acquisition Time: Clutching was faster than clutching-free acquisition, with M = 2.807, SD = .813 versus M = 3.83, SD = .723.This was a significant main effect of clutching type, F(1,207) = 96.214, p < .001, η2p = .32.
  • 5.2 Initial Acquisition Time: The technique-by-clutching interaction was significant, F(5,207) = 24.948, p < .001, η2p = .38.uniAngle and uniDepth were faster clutching-free, whereas biSemi and biDistance were slower; uniMicro and uniSemi showed no significant difference.
  • 5.2 Initial Acquisition Time: biDistance had the lowest initial acquisition-time mean at M = .299, SD = .214, while uniMicro had the highest at M = 1.734, SD = .835.biDistance, uniAngle, and uniDepth were significantly faster than biSemi, uniSemi, and uniMicro.
  • 5.2 Initial Acquisition Time: All techniques were faster clutching-free for initial acquisition, with uniMicro showing the largest degradation and biSemi the only nonsignificant difference.uniMicro’s clutching difference was Δ = 1.768, while uniSemi’s was Δ = 1.374.

5.4 Error Rate (Figure 5 (d))

Error rate and attempt count varied significantly by technique and interacted with clutching status. biDistance produced the lowest error and fewest attempts, whereas uniMicro showed the highest error and attempt count.

  • 5.4 Error Rate: Error rate showed a significant technique effect, F(5,207) = 16.970, p < .001, η2p = .29.biDistance had the lowest error rate at M = 3.165, SD = 1.076, while uniMicro had the highest at M = 5.880, SD = 2.760.
  • 5.4 Error Rate: Error rate showed a significant technique-by-clutching interaction, F(5,207) = 6.200, p < .001, η2p = .13.uniMicro had the largest clutching difference, whereas uniSemi showed the smallest nonsignificant variation.
  • 5.4 Error Rate: Attempt count showed a significant technique effect, F(5,207) = 37.010, p < .001, η2p = .47.biDistance required the fewest attempts at M = 1.162, SD = .305, while uniMicro required the most at M = 2.098, SD = .434.
  • 5.4 Error Rate: All techniques required fewer attempts clutching-free, except that uniAngle and uniSemi showed no significant clutching difference.The technique-by-clutching interaction was significant, F(5,207) = 5.654, p < .001, η2p = .12.

5.6 Physical Hand Movement (Figure 5 (f))

Technique and clutching condition both shaped physical hand movement. uniMicro and biSemi required the least movement overall, while clutching-free operation particularly reduced movement for uniAngle and uniDepth.

  • uniMicro (M = .150) and biSemi (M = .197) exhibited the lowest hand movement, followed by biDistance, uniSemi, uniAngle, and uniDepth.The technique effect was significant, F(5,207) = 41.650, p < .001, η2p = .50.
  • uniAngle and uniDepth showed substantially reduced hand movement in clutching-free status, with differences of .377 and .435, respectively.Both effects were significant at p < .001.
  • uniMicro and uniSemi showed smaller clutching-condition differences of .001, while biDistance and biSemi increased hand movement in clutching-free status by .114 and .097.The reported differences were statistically significant, with p values from .002 to < .001.
  • Technique significantly affected NASA-TLX mental demand, effort, and frustration.The reported chi-square statistics were 57.697, 57.393, and 60.083, respectively, all with p < .001.
  • SUS scores ranged from 52.81 for uniSemi-clutching to 88.59 for biDistance-clutching, with eight of twelve combinations exceeding 68 points.biDistance and uniDepth demonstrated excellent usability, while uniSemi had the lowest scores in both conditions.

5.8 Ranking (Figure 7) and Qualitative Feedback

biDistance and uniDepth were the most preferred techniques across clutching conditions, while clutching-free operation improved rankings for most techniques. uniSemi was consistently least preferred.

  • uniSemi was consistently least preferred, with average ranks of 8.62–9.29 and the most bottom-three rankings.This pattern held across both clutching conditions.
  • Clutching-free operation increased preference rankings across most techniques, especially uniDepth and uniMicro.uniDepth’s top-preference ratio rose from 33.33% to 42.86%, while uniMicro’s top-preference ratio doubled.
  • Figure 6 reports experience measures on 7-point scales and SUS scores on 100-point scales, with 95% confidence intervals.Asterisks denote statistical significance between clutching conditions.
  • Figure 7 presents participants’ preference rankings across techniques and clutching conditions.

6 DISCUSSION

Clutching effectiveness depended on mapping structure: isomorphic techniques benefited from clutching, whereas non-isomorphic techniques performed better without it. User preference also reflected both gesture affordance and physical effort.

  • 6.1 Mapping structure as the determinant of clutching effectiveness: Clutching produced faster completion times and fewer attempts for isomorphic biSemi and biDistance techniques.Their direct body-state-to-object-size correspondence provided a persistent physical reference frame for pausing and resuming manipulation.
  • 6.1 Mapping structure as the determinant of clutching effectiveness: uniSemi showed no significant clutching-condition difference, because its threshold-based mode trigger proved unreliable during repeated scaling.The finding suggests clutching benefits isomorphic mappings only when activation is simple and immediate.
  • 6.1 Mapping structure as the determinant of clutching effectiveness: Clutching-free operation improved completion time and reduced hand movement for non-isomorphic uniDepth, uniMicro, and uniAngle.These mappings lack proprioceptive grounding and instead rely on users developing a learned sense of movement-to-scale correspondence.
  • 6.2 Gesture Affordance and Technique Preference: uniDepth was judged most natural because pulling the hand aligns with the embodied idea of bringing an object closer to enlarge it.This affordance alignment can reduce the cognitive gulf of execution for scaling.
  • 6.2 Gesture Affordance and Technique Preference: uniMicro minimized hand movement but produced more attempts and relatively high error rates because its high gain amplified unintended motions during fine-tuning.Small thumb movements generated disproportionately large scale changes, causing overshoot and corrective adjustments.
  • 6.2 Gesture Affordance and Technique Preference: biSemi combined accurate isomorphic finger-span control with low fatigue from its constrained finger-motion range.This combination avoided the usual precision–comfort trade-off described for full-hand bimanual scaling.
  • Design implications: uniMicro with clutching-free operation is positioned for repeated scaling because additional attempts do not significantly increase physical effort.The paper reports less hand movement and overall physical demand than biDistance and uniDepth.
  • Design implications: uniSemi and biSemi with clutching are positioned for high-precision tasks requiring strong physical ownership and spatial awareness.Their isomorphic mappings provide proprioceptive grounding for direct coupling to the scaled object.

7 LIMITATION & FUTURE WORK

The study’s multitasking claims remain provisional because scaling and translation were evaluated sequentially, while rotation and genuine concurrent interaction were not directly tested. The evaluation also used a simplified scene, bounded magnification, and a limited participant pool.

  • Scaling and translation were evaluated sequentially, so concurrent scaling with translation and rotation remains unverified.Future studies should test more dynamic XR settings to assess the approach’s true multitasking efficiency.
  • The study evaluated only scale and translation tasks, bounded magnification at 3×, and used a single interactable object in a simplified scene.These choices may not reflect applications involving repeated scaling, multiple objects, or cluttered environments.
  • Concurrent multitasking was implied as a benefit of unimanual input but was not directly evaluated.The paper identifies empirical verification as future work.
  • The participant pool consisted entirely of right-handed users, was mostly male, and largely had high prior VR experience.Fifteen of twenty participants were male.

8 CONCLUSION

UniScale addresses bimanual scaling’s structural limitations and physical fatigue in Gaze+Pinch XR environments. Its findings show that clutching should match the mapping strategy, while users may accept minor precision compromises for reduced fatigue and improved multitasking.

  • UniScale is a suite of unimanual gesture mapping strategies for Gaze+Pinch XR scaling.
  • Isomorphic mappings benefit from discrete clutching, whereas rate-based mappings excel with continuous, clutching-free control.
  • Users accepted minor precision compromises in exchange for significantly reduced physical fatigue.
  • Freeing the dominant hand enables more efficient spatial multitasking in XR.
Loading 2609.00500v1…