VR Sickness Versus VR Presence: A Statistical Prediction Model.

Area of Science:

• Virtual Reality
• Human-Computer Interaction
• Immersive Technologies

Background:

• Virtual reality (VR) sickness and sense of presence are key to user experience, yet methods to predict and optimize their trade-offs are lacking.
• Existing VR sickness assessment (VRSA) and VR presence assessment (VRPA) studies often use limited probes, hindering applicability to diverse real-world VR content.
• A comprehensive understanding of the relationship between VRSA and VRPA is crucial for enhancing VR immersion.

Purpose of the Study:

• To address the lack of definitive research on predicting and optimizing VR sickness and presence.
• To develop a predictive model for VR sickness and presence using a novel, large-scale VR dataset.
• To explore the interplay between VR sickness and presence in general VR environments.

Main Methods:

• Construction of a dedicated VR sickness/presence (VR-SP) database with 100 VR videos and subjective ratings.
• Development of a statistical model using spatio-temporal and rotational frame difference maps to predict VR sickness.
• Design of a novel motion feature correlating instantaneous and averaged temporal features, augmented with visual activity and content features for presence prediction.

Main Results:

• The developed statistical model accurately predicts VR sickness with 90% accuracy using the VR-SP dataset.
• The aggregate VR-SP model demonstrates a 75% accuracy in predicting VR presence.
• The study successfully explored the relationship between VRSA and VRPA using the new resource.

Conclusions:

• The VR-SP database and predictive model offer a significant advancement in understanding and managing VR user experience.
• The findings enable better prediction and potential optimization of the balance between VR sickness and presence.
• This research paves the way for more immersive and comfortable virtual reality applications.