A physically-based BSDF for modeling the appearance of paper
- Author(s): Papas, Marios
- et al.
In this thesis a novel physically based appearance model which utilizes both microfacet and diffusion theory is introduced. Although the model could potentially be used for a large variety of complex materials that interact with light that scatters through their surface and sub- surface a more needed and complex application was decided, paper. Paper, which is simply assumed to have diffuse characteristics by many, is actually a rough, complex material consisting of thin layers of cellulose pulp and smooth finishes. These layers give paper a unique front and back surface with each side varying in roughness and levels of gloss. Through the measured data, paper was found to exhibit specular reflection, retroreflection, sheen at grazing angles, and subsurface scattering, all of which is accounted for in the novel appearance model. The Paper BSDF successfully combines, in a physically correct manner, both microfacet theory and as well as scattering and diffusion theory. All of the underlying modified models used as building blocks, share the same parameters and light's interaction with the material is naturally predicted. The model's accuracy was verified through BSDF measurements acquired using the UCSD Hemispherical Gantry with the aid of a non-linear constrained SQP algorithm. All of the parameters of the model have physical meaning and can be easily altered to change the appearance of the material modeled in a predictable manner which provides a significant advantage over other fitting methods such as Lafortune lobes or Wavelets. Although the model proposed is a BSDF it has a direct correlation with a Bidirectional Sub-surface Scattering Distribution Function (BSSDF) which can be used with the same parameters to render the desired material. This essentially provides an efficient way to measure materials that exhibit sub-surface scattering properties as a BSDF and then with the extracted parameters the BSSDF can be used for rendering. The results show that the model can efficiently account for various paper types, with just changes in parameters, varying from matte to photographic glossy coated paper. Finally it is explained how the physically based nature of the model can be used to successfully reduce the search space of it physically meaningful parameters, in order to obtain faster convergence of the fitting algorithm