Home › Topics › 3D Data Modelling › Bidirectional Texture Function Modelling

Non-linear Reflectance Model for Bidirectional Texture Function Synthesis

   Jiří Filip    Michal Haindl    Petr Somol
Abstract: 
A rough texture modelling involve huge image dataset - the Bidirectional Texture Function (BTF). This texture function is 6-dimensional depending on planar texture coordinates as well as on view and illumination position. The main contribution of this work is a new non-linear reflectance model which enable to synthetise BTF of a rough materials with complicated anisotropic properties. The model is based on modified and extended Lafortune reflectance model computed per each texel. The extension consist in adding a few spectral parameters for each BTF image which are linearly estimated according to original data in second estimation step. A model parameters are computed for every surface reflectance field contained in original BFT data. The final memory BTF data storage demands are with using of this technique reduced in ratio 1:15 when the synthetised images are almost indiscernible from originals. The method is universal, robust and easily implementable in a graphical hardware.

Car interior covered by four different BTF wood,foil,fabric and leather approximated by means of proposed polynomial extension of Lafortune model.

Polynomial Extension of Lafortune model Polynomial Extension of Lafortune model
All BTF data by courtesy of R. Klein, Bonn University.
Mercedes Class-C 3D model by courtesy of DaimlerChrysler.
BTF data enlarged using image quilting technique.

BTF data: Knitted wool - rectified to head on view position from elevation angle 60 degrees and azimutal angle 54 degrees. Animated size 128x128 pixels. Red dot = light, blue dot = camera.

Standard one-lobe Lafortune model (left = original data, right = synthetised results).


Modified one-lobe Lafortune model (left = original data, right = synthetised results).



Comparison of two BTF materials (wool and proposte)
view & illumination positions
original measured BTF data
one-lobe Lafortune model
proposed modification


Comparison of reconstruction error (MAE) for BTF materials (wool,foil02,fabric02,wood02). The red graph represents one-lobe lafortune model only while the blue graph represents one-lobe lafortune model with proposed extension.



Example of two BTF materials (foil02, wood02) mapped on car gearbox. For each texel on object surface dedicated reflectance model is used according to actual view and illumination position.

Original BTF measurement rendering

One lobe Lafortune model only


Proposed method result
(Move mouse over the image to get result of one-lobe Lafortune model without proposed modification)

One lobe Lafortune model only



Two examles of BTF mapped on car armrest. For each texel on object surface dedicated reflectance model is used according to actual view and illumination position.

Leather02 material - move mouse over the image to get result of one-lobe Lafortune model without proposed modification.

One lobe Lafortune model only



Fabric02 material - move mouse over the image to get result of one-lobe Lafortune model without proposed modification.

One lobe Lafortune model only

All BTF data by courtesy of R. Klein, Bonn University.
Mercedes Class-C 3D model by courtesy of DaimlerChrysler.




This work was partially supported by the European Community within the scope of the RealReflect project (IST-2001-34744) "Realtime visualization of complex reflectance behavior in virtual prototyping''.

Reference: 
Filip, J., and M. Haindl, "Non-linear reflectance model for Bidirectional Texture Function synthesis", Proceedings of the 17th IAPR International Conference on Pattern Recognition, Los Alamitos, IEEE, pp. 80-83, August, 2004.
Filip, J., and M. Haindl, "Efficient Image Based Bidirectional Texture Function Model", Texture 2005: Proceedings of 4th Internatinal Workshop on Texture Analysis and Synthesis, Edinburgh, Heriot-Watt University, pp. 7-12, October, 2005.