a photometric approach for estimating normals and tangents

43

Upload: armani

Post on 15-Jan-2016

30 views

Category:

Documents


0 download

DESCRIPTION

A Photometric Approach for Estimating Normals and Tangents. Input: Images under varying point lighting. Output: Estimate of surface orientation. Normal Field. Tangent Field. Related Work Lambertian photometric stereo [Woodham 1980]. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: A Photometric Approach for Estimating Normals and Tangents
Page 2: A Photometric Approach for Estimating Normals and Tangents

A Photometric Approach for Estimating Normals and

TangentsMichael Holroyd

University of Virginia

Jason LawrenceUniversity of

Virginia

Greg HumphreysUniversity of

Virginia

Todd ZicklerHarvard University

Page 3: A Photometric Approach for Estimating Normals and Tangents

. . .

Input: Images under varying point lighting

Page 4: A Photometric Approach for Estimating Normals and Tangents

Output: Estimate of surface orientation

Normal Field Tangent Field

Page 5: A Photometric Approach for Estimating Normals and Tangents

Related Work Lambertian photometric stereo

[Woodham 1980]

Discard specular highlight[Coleman and Jain 1982; Mallick et al. 2005]

[Klette et al. 1998]

Page 6: A Photometric Approach for Estimating Normals and Tangents

Fit to low-dimensional parametric models[Georghiades 2003; Goldman et al. 2005]

Fit non-parametric curves (isotropic only)[Alldrin et al. 2008]

Page 7: A Photometric Approach for Estimating Normals and Tangents

Locate mirror direction[Wang and Dana 2006; Chen et al. 2006; Ma et al. 2007; Francken et al. 2008; Nehab et al. 2008]

Image: Ma et al. 2007

Page 8: A Photometric Approach for Estimating Normals and Tangents

Our Approach

Page 9: A Photometric Approach for Estimating Normals and Tangents

Our Approach

Page 10: A Photometric Approach for Estimating Normals and Tangents

Our Approach

Page 11: A Photometric Approach for Estimating Normals and Tangents

Half-angle parameterization

2D slice of the BRDF (fixed view)

Our Approach

Page 12: A Photometric Approach for Estimating Normals and Tangents

Half-angle parameterization

2D slice of the BRDF (fixed view)

Our Approach

Page 13: A Photometric Approach for Estimating Normals and Tangents
Page 14: A Photometric Approach for Estimating Normals and Tangents
Page 15: A Photometric Approach for Estimating Normals and Tangents
Page 16: A Photometric Approach for Estimating Normals and Tangents

Analysis of microfacet-based models:

Page 17: A Photometric Approach for Estimating Normals and Tangents

Microfacet distributionAlmost all analytic and measured distributions exhibit these symmetries.

Analysis of microfacet-based models:

Images: Ngan et al. 2005

Page 18: A Photometric Approach for Estimating Normals and Tangents

Analysis of microfacet-based models:

1

-90° 90°

Fresnel TermWell approximated by , and is asymmetric only at grazing angles.

Page 19: A Photometric Approach for Estimating Normals and Tangents

Shadowing/MaskingSmooth and can be greatly simplified.

Analysis of microfacet-based models:

Shadowing/Masking functionfrom Wang et al. 2008

[Torrance1987; Ashikhmin et al. 2000;

Ngan et al. 2005; Wang et al. 2008]

Page 20: A Photometric Approach for Estimating Normals and Tangents

Restrict light positions

Page 21: A Photometric Approach for Estimating Normals and Tangents
Page 22: A Photometric Approach for Estimating Normals and Tangents

Symmetry Distance

Page 23: A Photometric Approach for Estimating Normals and Tangents

Symmetry Distance

Page 24: A Photometric Approach for Estimating Normals and Tangents

viewview

normal

Restriction of light positions

Page 25: A Photometric Approach for Estimating Normals and Tangents

view

normal

Restriction of light positions

Page 26: A Photometric Approach for Estimating Normals and Tangents

Restriction of light positions

view

normal

Page 27: A Photometric Approach for Estimating Normals and Tangents

Validation

Normal Error Tangent Error

BRDF from Ngan et al. 2005

Page 28: A Photometric Approach for Estimating Normals and Tangents

Normal Error Tangent Error

BRDF from Ngan et al. 2005

Validation

Page 29: A Photometric Approach for Estimating Normals and Tangents

Anisotropic Ward

PurpleSatin

YellowSatin

BrushedMetal

Measured

Ngan et al. 2005

Page 30: A Photometric Approach for Estimating Normals and Tangents

Error Analysis (Torrance-Sparrow)N

orm

al E

rror

(D

egre

es) Our approach

Photometric stereoSpecularity stereo

Diffuse Shiny

Page 31: A Photometric Approach for Estimating Normals and Tangents

Acquisition

Calibrated spherical gantry

1,500 1024x1024 HDR images

2.3 GB

45 minutes

Page 32: A Photometric Approach for Estimating Normals and Tangents

1. Reconstruct a continuous 2D slice of the BRDF at each pixel using barycentric interpolation of the original data.

2. Estimate by optimizing

3. Estimate by holding fixed and optimizing

Algorithm

Page 33: A Photometric Approach for Estimating Normals and Tangents

Implementation

Independent at each pixel

42x Dual 1.6 Ghz Opertons

10 minutes

(~7 hours for single machine)

Page 34: A Photometric Approach for Estimating Normals and Tangents
Page 35: A Photometric Approach for Estimating Normals and Tangents
Page 36: A Photometric Approach for Estimating Normals and Tangents
Page 37: A Photometric Approach for Estimating Normals and Tangents
Page 38: A Photometric Approach for Estimating Normals and Tangents
Page 39: A Photometric Approach for Estimating Normals and Tangents

Limitations

Interreflections

Page 40: A Photometric Approach for Estimating Normals and Tangents

Limitations: Sampling density

1,512 743 380 172

Page 41: A Photometric Approach for Estimating Normals and Tangents

Limitations

Non-symmetric microfacet distributions

Red velvet dataset from Ngan et al. [2005]

Page 42: A Photometric Approach for Estimating Normals and Tangents

Conclusion

Main advantages: General, does not rely on parametric model First technique to directly recover tangent field

Page 43: A Photometric Approach for Estimating Normals and Tangents

Acknowledgments Jiajun Zhu for help with data capture NSF CAREER award CCF-0747220 NSF grant CCF-0811493 NVIDIA Professor Partnership award Fellowship from the Sloan Foundation

Thank You