Reflectance Imaging: A Simple Approach to Capturing Surface Detail - PowerPoint PPT Presentation

About This Presentation
Title:

Reflectance Imaging: A Simple Approach to Capturing Surface Detail

Description:

Reflectance Imaging: A Simple Approach to Capturing Surface Detail – PowerPoint PPT presentation

Number of Views:45
Avg rating:3.0/5.0
Slides: 25
Provided by: tommalz
Category:

less

Transcript and Presenter's Notes

Title: Reflectance Imaging: A Simple Approach to Capturing Surface Detail


1
Reflectance Imaging A Simple Approach
to Capturing Surface Detail
Tom Malzbender
Hewlett-Packard Laboratories
2
Texture Mapping Catmull 74
Bump Mapping Blinn 78
  • Pro
  • Lighting
  • Variations
  • Pro
  • Photographic
  • input
  • Simplicity
  • Hardware Support
  • Con
  • Per-pixel lighting
  • computation
  • Filtering is Problematic
  • Procedural Synthesis
  • (Not image-based.)
  • Rushmeier 97
  • Con
  • Unrealistic Silhouettes
  • Static Lighting

3
PTM Demonstration
  • Advantages
  • Image based unlike bump mapping
  • Simpler to evaluate than bump mapping
  • Can leverage mip mapping

4
PTM Demonstration
Top Polynomial Texture Map
Bottom Conventional Texture Map
  • Advantages
  • Image based unlike Bump Mapping
  • Simpler to evaluate than Bump Mapping
  • Can leverage Mip Mapping

5
Acquiring PTMs Photographically
  • Fixed object, fixed camera.
  • Limited to Diffuse Objects.

Third acquisition device being fabricated
currently
6
Modeling Pixel Color Changes Directly
(u,v)
Light Direction Space
7
Modeling Pixel Color Changes Directly
(u,v)
Light Direction Space
8
Polynomial Texture Mapping
PTM Store RGB per pixel and Store polynomial
coefficients (a0-a5) per textel
R L R' G L G' B L B'
  • Why Polynomials?
  • Compact Representation
  • Consist solely of multiplies and adds.
  • Cheap to evaluate on both modern CPUs and VLSI

9
Light Direction Parametrization
u,v - texture coordinates a0-a5 - fitted
coefficients stored in texture map lu,lv -
projection of light direction into texture plane
lv
lu
lu,lv can be scan-converted without
normalization.
10
PTM Formats
Per Pixel Storage
Format
  • LRGB -
  • RGB -
  • ENC - Index to L.U.T. storing Polynomial
    Coefficients
  • ...


R,G,B
11
2D Applications
  • Enhancement of Cuneiform
  • Tablets w/ Zuckerman USC
  • PTMs for Short Image Sequences
  • PTMs for Depth of Focus Effects.

12
Surface Normal Extraction
Yields maximum surface brightness
13
Surface Normal Extraction
For a diffuse object, coordinates of (lu,lv) that
maximize luminance yield local surface normals.
Setting
yields
Providing a surface normal per textel
14
Specular Enhancement
Cuneiform tablet courtesy of Dr. Bruce Zuckerman
at U.S.C.
15
Diffuse Gain - a reflection transformation that
  • Keeps the surface normal fixed.
  • Increases the second derivative
  • of the reflectance function by g.

16
Light Direction Extrapolation
  • Input images are collected across a
  • hemisphere of light directions, i.e. -1lt
    lu,lv lt1
  • PTMs can be evaluated outside of the
    hemisphere,
  • ( lu,lv lt -1 or lu,lv gt 1 )

17
Forensic Applications
18
Java PTM viewer
  • Written by Clifford Lyon, Harvard University
  • Easy to add new datasets.
  • http//materialobjects.com/browser/PTMBrowser_1.h
    tml

19
Forensic Applications
20
National Gallery London
Light Arm
Joe Padfield
21
  • Basler A504kc 500 F/sec camera.
  • Real-time transfer to HP workstation.
  • Processing on GPU, Nvidia Gforce 7800 GTX.
  • 60 f/sec throughput.
  • Normal transforms (unsharp mask), Relighting.

Real Time Enhancement
22
PTM Object Movies - POM
  • Viewpoint Variation
  • Lighting Variation
  • Surface Enhancements
  • Currently lacking in view interpolation.

Collaboration w/ Alberto Proenca, Univ. Minho,
Portugal Cultural Heritage Imaging, California
Collection Archéologie du Musée de l'Hospice du
Grand St. Bernard
23
Highlight-based PTMs
Developed with Mark Mudge, Cultural Heritage
Imaging
V
N
L
f
z
Sx
1.0
x
24
Conclusions Web Tools
  • Reflectance functions are easy to collect and
    model.
  • Transforming reflectance functions can help
    visualization.
  • Tools are available at hpl.hp.com/ptm
  • sample PTMs
  • PTM viewer
  • Polynomial Fitter
  • PTM format document
Write a Comment
User Comments (0)
About PowerShow.com