Realistic Materials for Virtual Real-Time Environments
Abstract
Bidirectional Texture Functions (BTF) can be used to render realistic surfaces without the need of modeling details in geometry. For efficient, real-time rendering on current graphics hardware it is necessary to find good approximations for the BTF data, because the full data is too large to be rendered. We will compare two approximations, the Polynomial Texture Map (PTM) and the Per-Pixel-Lafortune-BRDF. To use larger samples, we use texture synthesis methods based on BTF analysis with the BTF samples and compare the results obtained with different settings and materials. We combine these approaches with a rendering method that provides real-time rendering on large surfaces, but does only need little more memory than for the small samples.
Stichwörter: appearance, BTF, Image Based Rendering, real-time rendering, texture synthesis
Bilder
![]() |
Paper herunterladen
Bibtex
@INPROCEEDINGS{bosserhoff-2003-realistic-materials, author = {Bo{\ss}erhoff, M. and Nicoll, A.}, title = {Realistic Materials for Virtual Real-Time Environments}, booktitle = {Central European Seminar on Computer Graphics for Students (CESCG 2003)}, year = {2003}, month = apr, keywords = {appearance, BTF, Image Based Rendering, real-time rendering, texture synthesis}, abstract = {Bidirectional Texture Functions (BTF) can be used to render realistic surfaces without the need of modeling details in geometry. For efficient, real-time rendering on current graphics hardware it is necessary to find good approximations for the BTF data, because the full data is too large to be rendered. We will compare two approximations, the Polynomial Texture Map (PTM) and the Per-Pixel-Lafortune-BRDF. To use larger samples, we use texture synthesis methods based on BTF analysis with the BTF samples and compare the results obtained with different settings and materials. We combine these approaches with a rendering method that provides real-time rendering on large surfaces, but does only need little more memory than for the small samples.} }