Efficient Resampling, Compression and Rendering of Metallic and Pearlescent Paint

M. Magnor, B. Rosenhahn, and H. Theisel (Editors)
In proceedings of Vision, Modeling, and Visualization, pages 11-18, Nov. 2009
 

Abstract

An interesting class of materials are metallic and pearlescent paints. They pose serious challenges to computer graphics since they have high frequencies in both angular and spatial domain, have a complex, pure stochastic spatial structure and angular dependent color shifts. Hybrid approaches that combine model and image-based representations have proven to faithfully reproduce metallic and pearlescent paint by compressing measured data. But until now the compression of the image based part is done with PCA based methods resulting in high amounts of data. One additional problem is the angular sampling of the measured data. Typical angular sampling densities of image based reflectance acquisition devices are very low compared to the solid angle, in which a metallic flake is visible, resulting in severe undersampling of the angular frequencies. We propose a simple method to measure these frequencies generated by the metallic flakes. Furthermore, we present a novel representation for metallic and pearlescent paint that allows resampling in angular domain corresponding to the flake properties and compression of measured image based paint data for high-quality paint rendering.

Images

Download Paper

Download Paper

Bibtex

@INPROCEEDINGS{rump-vmv-09,
     author = {Rump, Martin and Sarlette, Ralf and Klein, Reinhard},
     editor = {Magnor, M. and Rosenhahn, B. and Theisel, H.},
      pages = {11--18},
      title = {Efficient Resampling, Compression and Rendering of Metallic and Pearlescent Paint},
  booktitle = {Vision, Modeling, and Visualization},
       year = {2009},
      month = nov,
   abstract = {An interesting class of materials are metallic and
               pearlescent paints. They pose serious challenges
               to computer graphics since they have high frequencies
               in both angular and spatial domain, have a
               complex, pure stochastic spatial structure and angular
               dependent color shifts. Hybrid approaches
               that combine model and image-based representations
               have proven to faithfully reproduce metallic
               and pearlescent paint by compressing measured
               data. But until now the compression of the image
               based part is done with PCA based methods resulting
               in high amounts of data. One additional problem
               is the angular sampling of the measured data.
               Typical angular sampling densities of image based
               reflectance acquisition devices are very low compared
               to the solid angle, in which a metallic flake is
               visible, resulting in severe undersampling of the angular
               frequencies. We propose a simple method to
               measure these frequencies generated by the metallic
               flakes. Furthermore, we present a novel representation
               for metallic and pearlescent paint that allows
               resampling in angular domain corresponding to the
               flake properties and compression of measured image
               based paint data for high-quality paint rendering.}
}