A data-driven reflectance model is presented, which uses acquired reflectance data to represent BRDFs as dense sets of measurements. This approach allows interpolation and extrapolation in the space of acquired BRDFs to create new BRDFs. Each BRDF is treated as a high-dimensional vector, and both linear and non-linear dimensionality reduction techniques are applied to find a lower-dimensional representation. Users can define perceptually meaningful parameters to navigate the reduced BRDF space. The model also derives two novel sampling procedures that require fewer measurements than standard methods. Wavelet analysis is used to derive a common basis for reflectance functions and a non-uniform sampling pattern. The reflectance of materials can be represented as a linear combination of surface reflectance functions, enabling efficient measurement. The model is supported by extensive research and analysis, including the use of wavelets for efficient storage and measurement. The thesis also discusses the limitations of analytic models and the advantages of the data-driven approach in capturing realistic BRDFs. The model is validated through extensive experiments and comparisons with analytic models, demonstrating its effectiveness in capturing complex reflectance properties. The work contributes to the field of computer graphics by providing a flexible and efficient method for modeling surface reflectance.A data-driven reflectance model is presented, which uses acquired reflectance data to represent BRDFs as dense sets of measurements. This approach allows interpolation and extrapolation in the space of acquired BRDFs to create new BRDFs. Each BRDF is treated as a high-dimensional vector, and both linear and non-linear dimensionality reduction techniques are applied to find a lower-dimensional representation. Users can define perceptually meaningful parameters to navigate the reduced BRDF space. The model also derives two novel sampling procedures that require fewer measurements than standard methods. Wavelet analysis is used to derive a common basis for reflectance functions and a non-uniform sampling pattern. The reflectance of materials can be represented as a linear combination of surface reflectance functions, enabling efficient measurement. The model is supported by extensive research and analysis, including the use of wavelets for efficient storage and measurement. The thesis also discusses the limitations of analytic models and the advantages of the data-driven approach in capturing realistic BRDFs. The model is validated through extensive experiments and comparisons with analytic models, demonstrating its effectiveness in capturing complex reflectance properties. The work contributes to the field of computer graphics by providing a flexible and efficient method for modeling surface reflectance.