FNOPT: Resolution-Agnostic, Self-Supervised Cloth Simulation using Meta-Optimization with Fourier Neural Operators

2025-12-05

AI Summary

Existing neural cloth simulators exhibit limited generalization, struggling to simultaneously ensure cross-resolution consistency and high-fidelity motion detail preservation. To address this, we propose FNOpt—the first self-supervised cloth simulation framework that embeds the Fourier Neural Operator (FNO) into a meta-optimization architecture, formulating time integration as a differentiable optimization problem subject to physical constraints. FNOpt employs an FNO-parameterized meta-optimizer, a self-supervised physical loss, and a multi-scale rollout stabilization mechanism. Crucially, it enables zero-shot cross-resolution transfer without ground-truth supervision or retraining. In benchmark evaluations, FNOpt significantly outperforms state-of-the-art learning-based methods in both accuracy and robustness, particularly excelling at synthesizing fine-grained structural details such as wrinkles and folds.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Computer Vision: Learning & Optimization for CV
• Intelligent Robots: Learning & Optimization for ROB

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We present FNOpt, a self-supervised cloth simulation framework that formulates time integration as an optimization problem and trains a resolution-agnostic neural optimizer parameterized by a Fourier neural operator (FNO). Prior neural simulators often rely on extensive ground truth data or sacrifice fine-scale detail, and generalize poorly across resolutions and motion patterns. In contrast, FNOpt learns to simulate physically plausible cloth dynamics and achieves stable and accurate rollouts across diverse mesh resolutions and motion patterns without retraining. Trained only on a coarse grid with physics-based losses, FNOpt generalizes to finer resolutions, capturing fine-scale wrinkles and preserving rollout stability. Extensive evaluations on a benchmark cloth simulation dataset demonstrate that FNOpt outperforms prior learning-based approaches in out-of-distribution settings in both accuracy and robustness. These results position FNO-based meta-optimization as a compelling alternative to previous neural simulators for cloth, thus reducing the need for curated data and improving cross-resolution reliability.

Problem

• Simulates cloth dynamics without needing ground truth data
• Generalizes across mesh resolutions without retraining
• Captures fine-scale wrinkles while maintaining simulation stability

Innovation

• Self-supervised cloth simulation via meta-optimization
• Resolution-agnostic Fourier neural operator parameterization
• Generalizes to fine wrinkles without retraining