Quadtrees on the GPU

Abstract : We present a method to implement quadtrees on the GPU. It relies on linear trees, a pointer-free alternative to recursive trees, that represents nodes as bit codes. We explain how to update this data structure with node splitting and merging, and show how to render multiresolution, crack-free surfaces with frustum culling using hardware tessellation and a distance-based LOD selection criterion. Our implementation is both fast and lightweight, and runs asynchronously on the GPU, leaving CPU resources available for other tasks.
Document type :
Book sections
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-01530430
Contributor : Équipe Gestionnaire Des Publications Si Liris <>
Submitted on : Wednesday, May 31, 2017 - 4:57:53 PM
Last modification on : Thursday, November 1, 2018 - 1:19:43 AM

Identifiers

Citation

Jonathan Dupuy, Jean-Claude Iehl, Pierre Poulin. Quadtrees on the GPU. GPU Pro 5, A K Peters/CRC Press, pp.439-450, 2014, ⟨10.1201/b16721-33⟩. ⟨hal-01530430⟩

Share

Metrics

Record views

204