Abstract
Creating high-fidelity 3D environments for Virtual Reality (VR) is a resource-intensive process. This study investigates whether 360-degree panorama-based environments can serve as a viable, low-cost alternative to handcrafted 3D models without sacrificing the user's sense of presence. By comparing a meticulously modeled 3D room with a panorama-based replica in a stationary VR task, researchers found no statistically significant difference in immersion levels, suggesting a significant shortcut for developers of stationary VR experiences.
Introduction
The demand for realistic, immersive environments in VR often creates a development bottleneck. High-quality 3D modeling requires specialized artistry, significant time investments, and high computational overhead for rendering. For instance, the test environment used in this study required approximately 120 hours of expert labor to construct.
An emerging alternative is the use of 360-degree panoramas, which can be captured in seconds using consumer-grade hardware and are computationally inexpensive to render. However, the primary concern with this approach is whether replacing true 3D geometry with a flat, spherical image breaks the user's immersion. This research directly compares the "sense of presence" between these two methods to determine if the visual shortcut impacts the user experience.
Methodology
The researchers developed two visually identical VR meeting rooms to serve as the testing environments:
- Handcrafted 3D Environment: A full 3D model featuring realistic lighting, high-resolution textures, and complex geometry, representing 120 hours of development.
- Panorama-Based Environment: A spherical image viewer utilizing 5376 x 2688 resolution photographs taken from the exact seating position of the real-world room.
To ensure the panorama environment remained functional, the authors employed a hybrid approach: interactive 3D virtual objects (such as a projector screen and a calculator-like keypad) were superimposed over the 2D background.
User Study
The study involved 8 participants with no prior VR experience. Using an Oculus Quest 2 with hand tracking, participants were seated and tasked with a visual search exercise (e.g., counting specific items on the walls) to encourage 360-degree observation. Following the tasks, participants completed a subset of the Igroup Presence Questionnaire (IPQ) within the VR environment to rate their sense of "being there" on a 7-point Likert scale.
Key Results and Analysis
The study's primary finding was that there was no statistically significant difference in the sense of presence between the handcrafted 3D model and the panorama-based environment.
Quantitative Findings
Across all five IPQ metrics, one-way ANOVA tests yielded $p > 0.05$, indicating that users felt equally "present" in both environments. The data suggests that for stationary tasks, the brain does not necessarily require full 3D geometry to maintain the illusion of a physical space.
Qualitative Observations
Participants reported a high sense of realism in both scenarios. Interestingly, the illusion was so seamless that one participant failed to notice when the system transitioned from the 3D model to the 2D panorama. However, some users noted that the panorama felt slightly more pixelated, a result of stretching the 5.3K image across the entire viewing sphere.
Discussion: The "Near-Field" Illusion
A critical insight from this research is the effectiveness of the "near-field" illusion. By placing interactive 3D objects within the user's immediate reach, the system anchors the user in a 3D space. This interactive foreground effectively masks the static nature of the 2D background, allowing the panorama to feel like a deep, three-dimensional environment.
Limitations and Future Work
While promising, the study identifies several constraints:
- Sample Size: The small group of 8 participants limits the broader statistical power of the results.
- Lack of Parallax: Panoramas lack 6-Degrees-of-Freedom (6DoF) positional tracking. If a user moves their head significantly or attempts to walk, the perspective does not shift, breaking the illusion. This makes panoramas suitable only for stationary or teleportation-based navigation.
- Resolution Requirements: To match the crispness of rendered 3D models, future implementations likely require 8K resolution or higher.
Conclusion
This research demonstrates that 360-degree panoramas are a highly effective tool for VR developers. For experiences where the user remains relatively stationary, panoramas provide a sense of presence indistinguishable from expensive 3D models, offering a path toward faster and more accessible VR content creation.
Summary updated