Hierarchical Bitmask Implicit Grids for Efficient Point-in-Volume Queries on the GPU

Julius Ikkala, Tuomas Lauttia, Pekka Jääskeläinen, Markku Mäkitalo

2024

Abstract

We propose “Hierarchical Bitmask Implicit Grids”, a novel, memory-efficient spatial index data structure for querying bounding volumes based on a contained point, targeting real-time use cases on GPUs. Like grid structures based on 3D arrays, implicit grids allow for nearly array-like direct lookups of cells without traversal through a spatial tree structure. However, the space complexity of this structure is O(n) with respect to resolution as opposed to O(n 3 ), which allows for dramatically higher resolutions than would be feasible with a 3D array. We demonstrate the effectiveness of this data structure by applying it to two example use cases: light culling and decal rendering. We measure both cases with ray tracing and multi-view rendering. We show that with tens of thousands of entries, our data structure can be built in 0.1–0.2 milliseconds, being ∼ 2.9x faster than the compared state-of-the-art decal method and orders of magnitude faster than dense 3D arrays, while delivering at least similar or even up to doubled rendering performance.

Paper Citation

in Harvard Style

Ikkala J., Lauttia T., Jääskeläinen P. and Mäkitalo M. (2024). Hierarchical Bitmask Implicit Grids for Efficient Point-in-Volume Queries on the GPU. In Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 1: GRAPP; ISBN 978-989-758-679-8, SciTePress, pages 285-292. DOI: 10.5220/0012421600003660

in Bibtex Style

@conference{grapp24,
author={Julius Ikkala and Tuomas Lauttia and Pekka Jääskeläinen and Markku Mäkitalo},
title={Hierarchical Bitmask Implicit Grids for Efficient Point-in-Volume Queries on the GPU},
booktitle={Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 1: GRAPP},
year={2024},
pages={285-292},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012421600003660},
isbn={978-989-758-679-8},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 1: GRAPP
TI - Hierarchical Bitmask Implicit Grids for Efficient Point-in-Volume Queries on the GPU
SN - 978-989-758-679-8
AU - Ikkala J.
AU - Lauttia T.
AU - Jääskeläinen P.
AU - Mäkitalo M.
PY - 2024
SP - 285
EP - 292
DO - 10.5220/0012421600003660
PB - SciTePress