[Paper Review] Progressive tearing and cutting of soft-bodies in high-performance virtual reality

Manos Kamarianakis, Antonis Protopsaltis+3 more|arXiv (Cornell University)|Sep 18, 2022

3D Shape Modeling and Analysis4 citations

TL;DR

This paper presents a real-time, geometry-based framework for progressive tearing and cutting of rigged soft-body meshes in virtual reality, achieving sub-10ms performance on low-spec devices by leveraging particle systems layered over linear-blend skinned meshes. The method enables realistic, unconstrained surgical-grade cuts without pre-processing or FEM overhead, outperforming existing approaches in speed and flexibility for VR medical training applications.

ABSTRACT

We present an algorithm that allows a user within a virtual environment to perform real-time unconstrained cuts or consecutive tears, i.e., progressive, continuous fractures on a deformable rigged and soft-body mesh model in high-performance 10ms. In order to recreate realistic results for different physically-principled materials such as sponges, hard or soft tissues, we incorporate a novel soft-body deformation, via a particle system layered on-top of a linear-blend skinning model. Our framework allows the simulation of realistic, surgical-grade cuts and continuous tears, especially valuable in the context of medical VR training. In order to achieve high performance in VR, our algorithms are based on Euclidean geometric predicates on the rigged mesh, without requiring any specific model pre-processing. The contribution of this work lies on the fact that current frameworks supporting similar kinds of model tearing, either do not operate in high-performance real-time or only apply to predefined tears. The framework presented allows the user to freely cut or tear a 3D mesh model in a consecutive way, under 10ms, while preserving its soft-body behaviour and/or allowing further animation.

Motivation & Objective

• To enable real-time, unconstrained tearing and cutting of soft-body mesh models in high-performance virtual reality.
• To overcome the limitations of predefined or FEM-based cutting methods that are too slow or inflexible for mobile VR environments.
• To preserve soft-body deformation behavior and animation capabilities after tearing, ensuring realism in surgical simulations.
• To develop a framework compatible with mainstream game engines like Unity3D and Unreal Engine, suitable for deployment on low-spec mobile VR HMDs.
• To support collaborative and immersive VR training scenarios by enabling dynamic, user-driven interactions with deformable tissues.

Proposed method

• The framework uses a particle system layered on top of a linear-blend skinned mesh to simulate soft-body deformation, enabling realistic tissue response after tearing.
• Tearing is performed via pure geometric operations on the rigged mesh surface, avoiding expensive re-meshing or FEM solvers.
• The method computes bone weights and particle updates using efficient Euclidean geometric predicates, ensuring low computational overhead.
• Tear operations are decomposed into atomic steps: intersection detection, particle disconnection, mesh update, and particle state propagation.
• All operations are optimized for real-time execution within 10ms per frame, with minimal GPU/CPU load, enabling deployment on mobile VR devices.
• The system integrates seamlessly into game engines like Unity3D and is already available via the MAGES SDK.

Experimental results

Research questions

• RQ1 Can progressive, continuous tearing of soft-body meshes be achieved in real-time (≤10ms) within mobile VR environments without pre-processing or FEM overhead?
• RQ2 How can realistic soft-body deformation be preserved after tearing, especially in surgical simulation contexts?
• RQ3 Can a geometry-based approach outperform FEM or pre-defined animation methods in terms of performance and flexibility for VR applications?
• RQ4 What is the computational overhead of enabling unconstrained, user-driven cuts on complex models (e.g., heart, bunny) in real-time VR?
• RQ5 How does the method scale across models of varying complexity and mesh resolution in real-time VR rendering?

Key findings

• The framework achieves a total tear operation time of 3.25ms for a sphere, 11.19ms for a bunny, and 18.65ms for a heart model, all under the 10ms threshold for high-performance VR.
• The method outperforms state-of-the-art approaches like He et al. (2022), which required 52.77ms for a cuboid with 18,128 faces, while our method achieved 18.65ms for a similarly complex heart model.
• The particle update and disconnection steps are the most computationally intensive, with 2.63ms for particle disconnection on the heart model, still within real-time limits.
• The framework maintains minimal performance overhead during VR sessions, even with high-complexity models, due to the user’s natural mental preparation time between actions.
• The method supports dynamic soft-body behavior post-tearing, allowing further deformation and animation on the torn parts, crucial for realistic surgical simulation.
• The system is already integrated into the MAGES SDK and runs in Unity3D, demonstrating practical deployability in real-world VR medical training applications.