Barycentric Shaders: Art Directed Shading Using Control Images
Loading...
Date
2016
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
The Eurographics Association
Abstract
In this paper, we present Barycentric Shaders, a shading framework based on barycentric algebra, for the development of shader functions providing intuitive art-directed control. The framework guarantees that whatever style is desired will be consistently obtained regardless of the underlying rendering method and illumination model, since our shaders are used only to compute colors based on the incoming illumination. This property of our framework allows our shaders to be included in any rendering pipeline without major changes. We define barycentric operations over positive real colors to guarantee that results will always be positive real colors. With this formalization, we redefine shader functions as parametric functions that satisfy the partition of unity. This property supports an intuitive interaction mechanism for obtaining desired styles by guaranteeing that colors will always stay inside of the convex hull of a set of control colors. To obtain colored light effects, we extend our barycentric methods by allowing computation separately along each color channel, providing the convex hull property for each dimension independently. This results in a more relaxed rectangular box property without significantly changing visual style. This formalism can be particularly helpful to artists, who may not have mathematical training, by simplifying shader development to obtain a desired expressive style. This new approach also suggests a new rendering and shading architecture that provides a clear distinction between illumination and shading.
Description
@inproceedings{10.2312:exp.20161062,
booktitle = {Computational Aesthetics},
editor = {Angus Forbes and Lyn Bartram},
title = {{Barycentric Shaders: Art Directed Shading Using Control Images}},
author = {Akleman, Ergun and Liu, S. and House, Donald},
year = {2016},
publisher = {The Eurographics Association},
ISSN = {1816-0859},
ISBN = {978-3-03868-000-0},
DOI = {10.2312/exp.20161062}
}