To Slide or Not To Slide: Exploring Techniques for Comparing Immersive Videos

Paper Title

To Slide or Not to Slide: Exploring Techniques for Comparing Immersive Videos

Publication Info

• Topic area: Techniques for comparing immersive 360° videos in VR and 2D environments.
• Keywords: immersive videos, 360° video, virtual reality, video comparison, sliding concept, side-by-side, toggle, user study, ROI, spatial-temporal navigation.

Background and Problem

• Problem / challenge: Comparing immersive videos (IVs) is cognitively demanding due to their omnidirectional nature, and existing 2D video comparison techniques (e.g., toggle, side-by-side) do not directly translate to IVs. VR-based IV comparison faces challenges like memory strain and disorientation, while 2D projections distort spatial information.
• Significance: With IVs increasingly used in education, training, and entertainment, effective comparison techniques are critical for tasks like editing, quality evaluation, and production.
• Motivation and related work: Prior research has focused on single-video navigation and editing in VR or 2D video comparison techniques. However, there is a lack of systematic study on IV comparison, particularly techniques that combine flexibility and efficiency across VR and 2D modalities.

Solution

• Proposed approach: The paper adapts the "sliding" concept from 2D media comparison to immersive video comparison, enabling dynamic transitions between toggle and side-by-side views in both VR and 2D environments.
• Novelty:
  1. Extension of the sliding concept to immersive video comparison, resulting in two new techniques: SlideInVR and SlideIn2D.
  2. Implementation of five IV comparison techniques across VR and 2D modalities: SlideInVR, ToggleInVR, SlideIn2D, ToggleIn2D, and SideBySideIn2D.
  3. Empirical evaluation of these techniques through a user study (N = 20) covering four IV comparison tasks.
  4. Insights into how task type and modality shape user workflows, strategies, and preferences.
• Procedure and key techniques:
  1. Sliding concept: Combines toggle and side-by-side views with three core interactions—extend, slide, and swap—to adjust visible regions and layer order dynamically.
  2. Techniques:
     • SlideInVR: Supports sliding in VR with spherical-lune display areas.
     • SlideIn2D: Adapts sliding for 2D environments with partial overlays.
     • ToggleInVR and ToggleIn2D: Overlay-based toggling in VR and 2D.
     • SideBySideIn2D: Persistent side-by-side comparison in 2D.
  3. Features include ROI-based adjustments, minimaps, temporal navigation, and peek functionality for partial overlays.

Results

• Concrete findings:
  • SlideInVR and SlideIn2D were the most preferred techniques, valued for their flexibility and ability to switch between comparison approaches.
  • No significant differences in task accuracy across techniques (mean accuracy: 64–75%).
  • VR was perceived as more natural for navigation and effective for detecting quality differences but required higher physical effort and had a steeper learning curve.
• Advantage over baselines:
  • SlideInVR and SlideIn2D allowed dynamic allocation of space and supported both fine-grained and holistic comparisons, outperforming static techniques like ToggleIn2D and SideBySideIn2D in user preference.
  • SideBySideIn2D was effective for simple and low-workload tasks but lacked flexibility.
• Experiments / evaluation:
  • User study with 20 participants comparing five techniques across four task types: temporal occurrence, spatial-temporal motion, local visual differences, and global visual differences.
  • Metrics included task accuracy, workload (NASA-TLX), usability (UMUX-Lite), and user preferences.
• Limitations and future work:
  • Subjective interpretation of some tasks led to variability in accuracy.
  • Focused on monoscopic 360° IVs and pairwise comparisons; future work should explore stereoscopic IVs, multi-video comparisons, and domain-specific workflows.
  • Potential for integrating AI to automate ROI tracking and layout adjustments.

Summary

This paper introduces and evaluates five techniques for comparing immersive videos in VR and 2D, with a focus on the sliding concept that enables flexible transitions between toggle and side-by-side views. SlideInVR and SlideIn2D emerged as the most favored techniques due to their flexibility and support for diverse comparison tasks. While VR provided a more natural and immersive experience, it also imposed higher physical and cognitive demands. The findings highlight the importance of task-dependent strategies and suggest opportunities for refining interaction designs to improve usability and scalability for more complex scenarios.