im imag age ba base sed br d brdf de f desi sign
play

Im Imag age-ba base sed BR d BRDF de F desi sign Guide Image - PowerPoint PPT Presentation

Im Imag age-ba base sed BR d BRDF de F desi sign Guide Image Edited/Painted Image Create BRDF Paint Ezra Davis 1 , Weiqi Shi 1 , Hongzhi Wu 2 , Julie Dorsey 1 , and Holly Rushmeier 1 1 Yale University (United States) 2 Zhejiang University


  1. Im Imag age-ba base sed BR d BRDF de F desi sign Guide Image Edited/Painted Image Create BRDF Paint Ezra Davis 1 , Weiqi Shi 1 , Hongzhi Wu 2 , Julie Dorsey 1 , and Holly Rushmeier 1 1 Yale University (United States) 2 Zhejiang University (China)

  2. Problem Definition • Artists/Designers need to specify materials for rendering digital scenes or creating new physical products. We consider the subproblem of specifying bidirectional reflectance distribution functions (BRDFs). Virtual – Games/Film [1] Physical – Product Design [2]

  3. Problem Definition: Conventional methods • Conventional methods often do not provide much control for the artist • Can be limited by being defined by fixed analytic functions to control features such as highlights

  4. Problem Definition: Conventional methods Things you can’t model with simple approach: [3] [4]

  5. Problem Definition: Measured BRDFs • Richer varieties of materials can be captured from physical materials • Takes time to measure • Requires physical samples • Hard to edit by artists Measured materials in the MERL database [5]

  6. Problem Definition: Our Approach Guide Image Edited/Painted Image Create BRDF Paint

  7. Background: Current Design Systems

  8. Background: Current Design Systems Can be powerful and general, but extraordinarily complex and indirect! [6] [7]

  9. Previous Work: Gaussian Material Synthesis • Uses Disney’s “Principled” shader • Navigate parameter variations by example • Gaussian Material Synthesis Zsolnai-Feher et al. ACM TOG 2018

  10. Previous Work: BRDF-Shop Paint highlights onto BRDF • BRDF-Shop: creating physically correct • bidirectional reflectance distribution functions, Colbert et al. Computer Graphics and Applications, 2006

  11. Previous Work: The Lit Sphere • Use a painted sphere to shade an object by its normals • The Lit Sphere: A Model for Capturing NPR Shading from Art, Sloan et al., Graphics Interface, 2001

  12. Our image-based BRDF design system Dense BRDF Guide Sparse BRDF Painted Calculate Calculate Paint

  13. Our image-based BRDF design system

  14. Our System: Select object Allows design in context for the user • Allows associating normal/light/view • direction for calculations (Or Custom)

  15. Our System: Set light position • Light direction • Allows associating normal/light/view direction for calculations

  16. Our System: Download guide image Provide an image that can be • imported to any image editor

  17. Our System: Paint image Any type of editing allowed: Paint on image • Paint on an image layer • Adjust contrast • Set colors in different • ranges.

  18. Our System: Upload painted image

  19. BRDF format Incident Theta Incident Phi Reflected Theta Normal Reflected Phi Theta Phi

  20. Sparse BRDF Reconstruction Reconstruct Create Sparse from same angle BRDF

  21. Sparse BRDF Reconstruction Reconstruct from Create Sparse different angle BRDF

  22. Dense BRDF reconstruction Sparse BRDF Dense BRDF PCA Reconstruction

  23. Dense BRDF Reconstruction • Our method based on Nielsen et al. paper for reconstructing measured BRDFs from few samples On Optimal, Minimal BRDF Sampling for Reflectance Acquisition. Nielsen et al., Siggraph Asia 2015

  24. User Experiences • All of the people that used the interface would be willing to use this method (though one artist preferred traditional parametric BRDF creation) • Some users like resulting image more than their painting

  25. Limitations No textures Limited by BRDF tables in database (Spatially Varying BRDFs)

  26. Future work • Use larger database of measured BRDFs • Alternate reconstruction methods • Use natural images instead of just painted ones

  27. Future work • How to present lighting direction? • How is high dynamic range input (e.g. specular highlights) obtained from a low dynamic range image? • What other refinements and feedback should we offer? (e.g. “blurring” the BRDF)

  28. Image-based BRDF design

  29. Questions and Comments

  30. Image sources • [1] https://all3dp.com/1/best-20-3d-animation-software/ • [2] Photo-realistic Rendering of Metallic Car Paint from Image-Based Measurements. Rump et al., Eurographics 2008 • [3] https://jsouguan.en.made-in-china.com/product/fjdmsYwSkTcv/China- Magic-Color-Interference-Powder-Violet-Iridescent-Pearl-Pigment-for- Paint.html • [4] https://www.trurofabrics.com/shot-silk-taffeta-fabric-purple- orange.html • [5] A data-driven reflectance model. Matusik et al., Siggraph 2003 • [6] https://lesterbanks.com/2013/12/cinema-4d-creating-node-based- materials-and-shaders-using-cmnodes/ • [7] http://www.design-corps.co.uk/tag/slate-material-editor/

  31. BRDF format Incident Theta Incident Phi Reflected Theta Normal Reflected Phi Theta Phi

Recommend


More recommend