Honeycomb-Surfels: Compact Hexagonal Primitives for Efficient Scene Reconstruction (Hexels, arXiv 2610.37668)

Forum topic · 小凯 · 2026-10-28

Summary

Hexels are a new planar hexagonal surface primitive with a built-in first-order spatial color field, proposed by researchers from Johns Hopkins (Prakhar Kaushik, Soumava Paul, Alan Yuille) for efficient 3D scene representation, novel view synthesis, and 2D image representation. Unlike Gaussian splatting, Hexels exploit locally planar regions with smoothly varying appearance using fewer, larger primitives, with hexagonal support offering better area and boundary efficiency than quadrilateral domains. Across 34 scenes, memory-matched Hexels improve PSNR over 2DGS by 4.53–7.38 dB while requiring lower training VRAM. Per-primitive comparisons show up to 4.74 dB higher PSNR with 2.31x faster rendering FPS. For surface reconstruction, Hexels reduce error by 28.9% on 15 DTU scans and achieve the highest mean F1 (34.38% vs 31.72% for 2DGS) on 6 Tanks and Temples scenes. Hexels also outperform Image-GS for 2D image representation, pushing the Pareto frontier of the quality-efficiency trade-off.

Paper Overview

Field: Computer Vision / 3D Reconstruction
Authors: Prakhar Kaushik, Soumava Paul, Alan Yuille
arXiv: 2610.37668

Motivation

Method

Key Results

Across 34 scenes, Hexels improve PSNR over memory-matched 2DGS by 4.53–7.38 dB across datasets, with lower training VRAM.
Per-primitive ablation: up to 4.74 dB better than 2DGS, with lower training VRAM and 2.31x faster FPS.
Surface reconstruction: 28.9% error reduction compared with memory-matched 2DGS across 15 DTU scans.
On 6 Tanks and Temples surface scenes, Hexels achieve the highest mean F1 among compared methods (34.38% vs 31.72% for 2DGS).
As a general primitive, Hexels also improve 2D image representation over Image-GS.
Hexels push the Pareto frontier on the quality-efficiency trade-off of surface-aligned primitives.

Original Abstract

Novel view synthesis and 3D reconstruction have advanced rapidly with radiance fields and splatting, yet their memory and computational demands limit their use in resource-constrained settings. We introduce Hexels: novel planar hexagonal surface elements with a first-order spatial color field for efficient 3D scene and 2D image representation. Hexels exploit locally planar regions with smoothly varying appearance using fewer, larger primitives. Their hexagonal support provides an area- and boundary-efficient domain, while the local color field captures appearance variation within each primitive. Across 34 scenes, Hexels improve PSNR over memory-matched 2DGS by 4.53-7.38 dB across datasets while requiring lower training VRAM. We further ablate Hexels by showing that per primitive, Hexels outperform 2DGS by upto 4.74 dB while using lower training VRAM and having 2.31 times faster FPS. Even in unconstrained settings, Hexels offers complementary quality-memory trade-offs. For efficient 3D surface reconstruction, Hexels reduce error by 28.9% compared with memory-matched 2DGS across 15 DTU scans. On 6 Tanks and Temples surface scenes, Hexels achieves the highest mean F1 among compared methods (34.38%, versus 31.72% for 2DGS). We further also show that Hexels, as a general primitive, improves 2D image representation over Image-GS. Hexels push the Pareto frontier on the quality-efficiency trade-off of surface-aligned primitives.

Link: https://arxiv.org/abs/2610.37668