← All resources

MeshSplatting: Differentiable Rendering with Opaque Meshes

Auto-discovered

Paper Advanced

Jan Held, Renaud Vandeghen, Sanghyun Son, Daniel Rebain, Matheus Gadelha, Yi Zhou, Anthony Cioppa, Ming C. Lin, Marc Van Droogenbroeck, Andrea Tagliasacchi

Visit resource ↗

Description

MeshSplatting: Differentiable Rendering with Opaque Meshes presents a mesh-based novel view synthesis method by Jan Held and collaborators that jointly optimizes geometry and appearance through differentiable rendering for AR/VR and game-engine mesh pipelines, unlike point-based 3D Gaussian Splatting.

It covers optimizing a semi-transparent triangle soup, restoring connectivity with restricted Delaunay triangulation, then fine-tuning toward opaque triangles. The page reports Mip-NeRF360 gains versus MiLo, faster training and lower memory, plus Unity and Blender demos and downloadable PLY assets.

It is for advanced researchers and engine programmers in neural rendering who need mesh outputs. Familiarity with 3DGS and differentiable rendering helps.

Access is a free project site with abstract, stage videos, comparisons, and code or asset links. A browser viewer is listed as coming soon.

It is useful when reconstructed scenes must export as opaque meshes for standard engines rather than stay Gaussian-only. Link last verified accessible on 2026-09-29. Link last verified accessible on 2026-09-28.