From Triangles to Gradients: Exploring DMesh++
/ 1 min read
Table of Contents
Overview
A brief exploration of DMesh++, a differentiable meshing framework connecting geometric optimization with gradient-based learning. Developed during the MIT Summer Geometry Initiative (SGI 2025) in collaboration with Adobe Research, this article examines how gradients flow through mesh connectivity and compares Minimum-Ball and Weighted Delaunay Triangulation methods for improved stability and accuracy.
Authored by Gabriel Isaac Alonso Serrato
in collaboration with Amruth Srivathsan