preprint, 2026 [ pdf (73mb) / more coming soon... ] Volumetric operators are powerful, but are ordinarily too expensive for large-scale geometry processing. We introduce an efficient and robust Monte Carlo method for estimating volumetric operators and use them to train a mesh-based contrastive model. 
ACM Transactions on Graphics (Proceedings of SIGGRAPH), 2026 [ project page / pdf (78mb) / arXiv / code / bibtex ] Generative modeling on surfaces requires delicate care due to irregular triangulations. We employ Matérn noise (as opposed to iid Gaussian noise) to enable triangulation-agnostic flow matching on meshes. 
ACM Transactions on Graphics (Proceedings of SIGGRAPH Asia), 2025 [ project page / pdf (64mb) / arXiv / code / bibtex ] A new neural architecture for learning on surfaces that overcomes several key deficiencies in existing methods. We demonstrate its ability to learn deformations of detailed, high-resolution meshes, as well as semantic segmentation. 
ACM Transactions on Graphics (Proceedings of SIGGRAPH), 2024 [ project page / pdf (34mb) / arXiv / code / bibtex ] A data augmentation strategy that enables diffusion models to smoothly interpolate between disjoint data modes. We train a diffusion model to blend multiple types of procedural noise patterns, even in the absence of "in-between" training data. 
SIGGRAPH Asia, 2023 (Conference Track) [ project page / pdf (30mb) / arXiv / bibtex ] What is the smoothest subspace that spans a set of points in latent space? We optimize smooth parametrizations of such subspaces in 3D generative models and use them to explore continuous variations of meshes. 
The College Mathematics Journal, 2019 [ pdf (1mb) / journal / bibtex ] The probability density function and moments (OEIS A279055) of the area of stochastically generated inscribed geometry are derived. Preliminary findings were presented at TUMC 2017.
|
