AppWand: Editing Measured Materials using Appearance-Driven Optimization
ACM Transactions on Graphics (Proc. SIGGRAPH), August 2007
Fabio Pellacini, Jason Lawrence
Abstract
We investigate a new approach to editing spatially- and temporallyvarying measured materials that adopts a stroke-based workflow. In our system, a user specifies a small number of editing constraints with a 3-D painting interface which are smoothly propagated to the entire dataset through an optimization that enforces similar edits are applied to areas with similar appearance. The sparse nature of this appearance-driven optimization permits the use of efficient solvers, allowing the designer to interactively refine the constraints. We have found this approach supports specifying a wide range of complex edits that would not be easy with existing techniques which present the user with a fixed segmentation of the data. Furthermore, it is independent of the underlying reflectance model and we show edits to both analytic and non-parametric representations in examples from several material databases.