R. Hegedus and A. Lucat and J. Redon and R. Pacanowski 60° June 22 nd Antoine LUCAT
Problematic How to qualify a BRDF measurement ? Graphics Photo vs. Rendering Qualitative error June 22 nd Antoine LUCAT 1
Problematic How to qualify a BRDF measurement ? Graphics Two setups = two different BRDFs [Obein15] Photo vs. Rendering Qualitative error June 22 nd Antoine LUCAT 1
Problematic How to qualify a BRDF measurement ? Optics Graphics 30° Photo vs. Rendering Qualitative error Quantitative error June 22 nd Antoine LUCAT 1
Problematic How to qualify a BRDF measurement ? Optics 30° Aim: - Create a database with quantified uncertainties - Material characterization Quantitative error June 22 nd Antoine LUCAT 1
The Setup Sphere Light Camera Inspired by [Matusik03] and [Marschner99] June 22 nd Antoine LUCAT 2
The Setup Sphere Light → 1 BRDF configuration : 1 pixel + 1 rotation Camera → Isotropic BRDF only Inspired by [Matusik03] and [Marschner99] June 22 nd Antoine LUCAT 2
Measurement Principle Raw image Real setup June 22 nd Antoine LUCAT 3
Measurement Principle Virtual setup June 22 nd Antoine LUCAT 3
Measurement Principle Reconstruction Raytracing Measured BRDF June 22 nd Antoine LUCAT 3
Uncertainties Propagation Initial error Final error ? June 22 nd Antoine LUCAT 4
Uncertainties Propagation Initial error Final error No correlation Perfect correlation June 22 nd Antoine LUCAT 4
Uncertainties Propagation +error +error +error +error Initial error Initial error Final error No correlation Perfect correlation June 22 nd Antoine LUCAT 4
Uncertainties Propagation +error +error +error +error Initial error Final error No correlation Perfect correlation June 22 nd Antoine LUCAT 4
Results June 22 nd Antoine LUCAT
Data & Resolution Specular Mirror Copper Teflon Spectralon Lambert. Standart June 22 nd Antoine LUCAT 5
Data & Resolution Specular 30° 60° Mirror Copper Teflon Spectralon Lambert. Standart June 22 nd Antoine LUCAT 5
Data & Resolution Specular Position uncertainty Mirror Value uncertainty Copper Teflon Spectralon Lambert. Standart June 22 nd Antoine LUCAT 5
Data & Resolution Specular Mirror Range ( , , ) : (0-85°, 0-90°, 0-360°) ~100 000 000 points for 1° rotation step Copper Uncertainty model results : Teflon Angular resolution : 0.1° → 12° % (normal to grazing view angle) Value uncertainty : 3% → 100% Spectralon Lambert. Standart June 22 nd Antoine LUCAT 5
Uncertainty Analysis View elevation Light elevation Azimuthal angle June 22 nd Antoine LUCAT 6
Uncertainty Analysis View elevation Light elevation Azimuthal angle June 22 nd Antoine LUCAT 6
Uncertainty Analysis View elevation Light elevation Azimuthal angle = 0 June 22 nd Antoine LUCAT 6
Uncertainty Analysis View elevation Light elevation Azimuthal angle = 0 Light independent June 22 nd Antoine LUCAT 6
Uncertainty Analysis View elevation Light elevation Azimuthal angle = 0 = 0 = 0 = 0 = 0 Light independent June 22 nd Antoine LUCAT 6
Uncertainty Analysis = 0 = 0 10 3 = 0 = 0 = 0 Light independent Unreliable grazing angle measurements →3 orders of magnitude June 22 nd Antoine LUCAT 6
Uncertainty Analysis = 0 = 0 = 0 = 0 = 0 Light independent Unreliable grazing angle measurements →3 orders of magnitude Main source of error : → focal and light uncertainties June 22 nd Antoine LUCAT 6
Future Work June 22 nd Antoine LUCAT
Issues and Future Work Non-uniformities : − White light speckle June 22 nd Antoine LUCAT 7
Issues and Future Work Non-uniformities : − White light speckle Without roughness June 22 nd Antoine LUCAT 7
Issues and Future Work Non-uniformities : − White light speckle with roughness → Surface diffraction June 22 nd Antoine LUCAT 7
Issues and Future Work Non-uniformities : − White light speckle − Light field June 22 nd Antoine LUCAT 8
Issues and Future Work Beam intensity Non-uniformities : − White light speckle − Light field Light stains June 22 nd Antoine LUCAT 8
Issues and Future Work Non-uniformities : − White light speckle − Light field Convolution : − Aperture diffraction → Lens diffraction June 22 nd Antoine LUCAT 9
Issues and Future Work Non-uniformities : − White light speckle − Light field Convolution : − Aperture diffraction time of exposure dynamic range f f o - e d a r T DOF f-number aberrations → Lens diffraction June 22 nd Antoine LUCAT 9
Conclusion BRDF measurement with quantified uncertainties Absolute definition of the setup's quality of measurement Possible comparison between different setups June 22 nd Antoine LUCAT 10
Thank you for your attention Data will be released on ALTA : http://alta.gforge.inria.fr/data.html June 22 nd Antoine LUCAT
Reference [Matusik03] Matusik W., Pfister H., Brand M., McMillan L. : Efficient isotropic BDRF measurement. In Proc. EGWR '03 (2003), pp. 241-247. [Marschner98] Marschner S.R., Westin S.H., Lafortune E.P.F., Torrance K.E., Greenberg D.P.: Image-Based BRDF Measurement Including Human Skin. In Proc. of 10th Eurographics Workshop on Rendering, p.139-152, June 1999. [Obein15] Obein G., Audenaert J., Ged G., Leloup F.B. : Metrological issues related to BDRF measurements around the specular direction in the particular case of glossy surfaces. In Proc. SPIE 9398, Measuring, Modeling, and Reproducing Material Appearance 2015, 93980D (March 13, 2015), doi:10.1117/12.2082518. June 22 nd Antoine LUCAT
Supplemental Depth-of-field ? Other results ? Formulas ? Order of errors by parameters (11) ? Videos ? June 22 nd Antoine LUCAT
Recommend
More recommend
