Multiple Kernels Noise for Improved Procedural Texturing - CNRS - Centre national de la recherche scientifique Accéder directement au contenu
Article Dans Une Revue The Visual Computer Année : 2012

Multiple Kernels Noise for Improved Procedural Texturing

Guillaume Gilet
DMI
Djamchid Ghazanfarpour
  • Fonction : Auteur
  • PersonId : 918071
DMI

Résumé

Procedural texturing is a well known method to synthesize details onto virtual surfaces directly during rendering. But the creation of such textures is often a long and painstaking task. This paper introduces a new noise function, called multiple kernels noise. It is characterized by an arbitrary energy distribution in spectral domain. Multiple kernels noise is obtained by adaptively decomposing a user-defined power spectral density (PSD) into rectangular regions. These are then associated to kernel functions used to compute noise values by sparse convolution. We show how multiple kernels noise (1) increases the variety of noisy procedural textures that can be modeled and (2) helps creating structured procedural textures by automatic extraction of noise characteristics from user-supplied samples.

Dates et versions

hal-00766595 , version 1 (18-12-2012)

Identifiants

Citer

Guillaume Gilet, Jean-Michel Dischler, Djamchid Ghazanfarpour. Multiple Kernels Noise for Improved Procedural Texturing. The Visual Computer, 2012, 28 (6-8), pp.679-689. ⟨10.1007/s00371-012-0711-2⟩. ⟨hal-00766595⟩

Collections

UNILIM CNRS XLIM
64 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More