A Differentiable Material Point Method Framework for Shape Morphing.

Area of Science:

• Computer Graphics
• Computational Physics
• Geometric Modeling

Background:

• Elastic shape morphing often struggles with complex topological changes.
• Existing methods may lack physical realism or temporal coherence.
• The material point method (MPM) offers a robust framework for simulating deformable objects with changing topology.

Purpose of the Study:

• To develop a novel, physically-based morphing technique for elastic shapes.
• To enable complex topological transitions within a unified simulation framework.
• To generate temporally coherent and detailed morphing sequences.

Main Methods:

• Leveraging the differentiable material point method (MPM).
• Implementing space-time control via per-particle deformation gradients.
• Employing a chained iterative optimization technique for sequence generation.

Main Results:

• Successfully demonstrated physically-based morphing for elastic shapes.
• Accommodated complex topology changes, including object fusion and fission.
• Produced detailed elastic deformations and coherent morphing sequences across challenging scenarios.

Conclusions:

• The proposed differentiable MPM approach provides a robust solution for physics-based elastic shape morphing.
• The method effectively handles dynamic topology changes with high fidelity.
• This technique advances the state-of-the-art in realistic shape transformation and animation.