Measurement and editing of metallic car paint BRDF

1
Measurement and editing of metallic car paint BRDF
Martin Rump Computer Graphics GroupUniversity
of Bonn, Germany
2
Outline

• Motivation
• Car paint composition and resulting visual
  effects
• Previous work
• Our approach
• Results

3
Motivation

• Measurement costly
• Sometimes only nice images interesting
• Design process want to try different paints

Measured
New
Editing
4
Outline

• Motivation
• Car paint composition and resulting visual
  effects
• Previous work
• Our approach
• Results

5
Car Paint Composition
Binder
Pearlescent/flip-flop paint
Simple paint
Metallic paint
Finish
Coating
Direction dependent color shift
Base paint
Primer surfacer
Electro- coat
Substrat
Car
Color pigments (30-100 nm)
Base color
Glittering
6
Outline

• Motivation
• Car paint composition and resulting visual
  effects
• Previous work
• Our approach
• Results

7
Previous Work (Simulation)

• Ershov et al. Rendering Pearlescent Appearance
  Based On Paint-Composition Modelling (2001)
• Multilayer model
• Calculate operators for sublayers
• Combine operators ? BRDF
• Choose colors for components

• Limitations
• many parameters
• no guarantee for physical correctness

8
Previous Work (Measurement)

• Günther et al. Efficient Acquisition and
  Realistic Rendering of Car Paint (2005)
• BRDF measurement
• Cook-Torrance model
• Realtime raytracing
• Editing by changing CT-parameters

• Limitations
• cannot model all colors flops
• no flakes

9
Previous Work (Measurement)

• Matusik et al. Efficient Isotropic BRDF
  Measurement (2003)
• Matusik et al. A Data-Driven Reflectance Model
  (2003)

• Limitations
• - no physical correctness
• no flakes

10
Previous Work (Measurement)

• Bidirectional Texture Function (BTF)

Limitation no straightforward editing possible
11
Summary

• Simulation of layers
• Too many (unknown) parameters
• No flakes
• BRDF measurement and fitting
• simple Cook-Torrance cannot model all effects
• No flakes
• BTF rendering
• Representation not suitable for editing

12
Outline

• Motivation
• Car paint composition and resulting visual
  effects
• Previous work
• Our approach
• Results

13
Hybrid approach

• Image-based representation of sparkles

BTF
Flake BTF
BRDF

• Model-based reflection with direction dependent
  colors

Direction dependent color
BRDF
Intensity BRDF
14
Our BRDF model

• Split color intensity
• Store color in direction dependent table
• Low-frequent ? coarse sampling sufficient
• 200KB table
• No restriction to mechanism of color shift

• Basic model

Sphere painted with color table
15
Separation of BTF

• Split BRDF and BTF part
• Take light/view direction for every single pixel
• Subtract fitted BRDF model
• Synthesize BTF part

-

16
Rendering

• Add together BRDF synthesised BTF
• Path Tracing (mentalray)
• High quality
• Tonemapping Use responsecurve from camera
• GPU shader (cgFX, nVidia 8800 GTX)
• Only simple and few light sources
• Environment mapping
• Tonemapping Reinhard et al. (2005)

17
Editing (1)

• Exchange color table
• Exchange flake BTF

18
Editing (2)

• Affine mapping of color plane
• Drag, scale, rotate points projected onto plane
• Preview color flop on sphere

Mxt
19
Editing (3)
DEMO
20
Outline

• Motivation
• Car paint composition and resulting visual
  effects
• Previous work
• Our approach
• Results

21
Results (path tracing)

• Painted sphere (BRDF only)

• Painted sphere (BRDFBTF)

22
Result Color edited
23
Result Color edited
24
Result Colorflakes changed
25
Conclusion

• Hybrid rendering approach combining BRDF BTF
• Direction dependent color table for arbitrary
  color shift
• Editing by affine transformation of color table
• Transformation in shader possible
• Editing by exchange of flake BTF

Questions?