Scene-Based Foveated Fluid Animation in Virtual Reality

Area of Science:

• Computer Graphics
• Virtual Reality
• Computational Fluid Dynamics

Background:

• Physically-based fluid animation in VR enhances user experience but demands significant computational resources.
• Existing methods often struggle with the high computational cost of realistic fluid simulations in VR.

Purpose of the Study:

• To propose a novel scene-based gaze-contingent fluid simulation system for VR.
• To enhance the efficiency and perceptual realism of fluid animations in VR through adaptive resource allocation.

Main Methods:

• Developed a highly adaptive fluid simulator integrated with a VR perceptual model.
• Utilized an eccentricity and curvature-dependent model to dynamically allocate computational resources.
• Incorporated user study findings on perceptual thresholds into a unified sizing function for particle resolution.

Main Results:

• Achieved up to a 3.62× performance improvement in VR fluid simulations.
• Maintained perceptually optimal particle resolution and spatio-temporal stability.
• Demonstrated superior perceptual realism and user experience validated by subjective evaluation.

Conclusions:

• The proposed gaze-contingent adaptive fluid simulation system significantly improves performance and visual quality in VR.
• Dynamic resource allocation based on human perception is effective for high-fidelity VR fluid animation.
• This approach offers a viable solution for computationally intensive fluid simulations in immersive environments.