NVIDIA Logo Spatial Intelligence Lab, Isaac, Hi-Fi Physics, Seattle Robotics Lab

Adaptive Volumetric Mechanical Property Fields Invariant to Resolution

ICML 2026
PDF Read Paper PDF Hi-res Paper (130MB) GitHub Code 🤗 Model/Data format_quote Citation settings VoMP v1

TL;DR: AdaVoMP generates high resolution fine-grained volumetric physics properties which can be used to produce interactive physically realistic worlds.
Opposed to VoMP v1, AdaVoMP can generate $16^3\times$ higher resolution properties, allow test-time scaling the resolution, and use a new generated adaptive volumetric structure.

Abstract

Accurate mechanical properties (or materials) Young's modulus ($E$), Poisson's ratio ($\nu$) and density ($\rho$) are essential for reliable physics simulation of digital worlds, but most 3D assets lack this information. We propose AdaVoMP, a method for predicting accurate dense spatially-varying ($E$, $\nu$, $\rho$) for input 3D objects across representations, improving the resolution, accuracy, and memory efficiency over the state-of-the-art. The foundation of our technique is a sparse and adaptive voxel structure SAV that efficiently represents both the input 3D shape and the material field output. We replace the fixed-voxel model of the most accurate prior method, VoMP, with a novel sparse transformer encoder-decoder model that learns to generate a unique SAV autoregressively for every input shape to represent its materials, achieving a resolution $16^3\times$ higher than prior art. Experiments show that AdaVoMP estimates more accurate volumetric properties, even with lesser test-time compute than all prior art. This allows us to convert high-resolution complex 3D objects into simulation-ready assets, resulting in realistic deformable simulations.

Method

SAV: Sparse Adaptive Voxels

SAV dino merge SAV material tree

SAV is a sparse adaptive voxel representation that we use to encode both the input 3D shape and the output spatially varying materials. By efficiently representing geometry and materials in adaptive structures, we can allocate less compute to predict areas of piecewise constant materials, common in everyday objects (the wooden surface, the metal frame), while recursively refining only the fine heterogeneous regions and boundaries.

Adaptive Geometry Transformer & Adaptive Material Generator

Method overview

Method Overview: input shape is encoded as AdaVoMP (top left), encoded (top right), and processed with our autoregressive Adaptive Material Generator (bottom), which is trained to output material field as SAV.

Acknowledgments

We thank Gilles Daviet for help in setting up some of the simulations. We thank Jean-Francois Lafleche for help with rendering. We thank Beau Perschall, Katherine Cheung for help in using the datasets. We thank Ruchik Thaker for help in releasing the code and data. We thank Andre Pradhana, Anka He Chen, Anita Hu, Charles Loop, Clement Fuji Tsang, Francis Williams, Hexu Zhao and Ken Museth for insightful discussions.

Citation

@inproceedings{dagli2026adavomp,
    title={Adaptive Volumetric Mechanical Property Fields Invariant to Resolution},
    author={Rishit Dagli and Donglai Xiang and Vismay Modi and Xuning Yang and Gavriel State and David I.W. Levin and Maria Shugrina},
    booktitle={Forty-third International Conference on Machine Learning},
    year={2026}
}