Parameterization locally looks like Euclidian space of collection - - PDF document

parameterization
SMART_READER_LITE
LIVE PREVIEW

Parameterization locally looks like Euclidian space of collection - - PDF document

2-Manifold What makes for a smooth manifold? Parameterization locally looks like Euclidian space of collection of charts Meshes Meshes mutually compatible ll ibl on their overlaps form an atlas Parameterizations are key CS


slide-1
SLIDE 1

Parameterization

  • f

Meshes

CS 176 Winter 2011

1

Meshes

2-Manifold

What makes for a smooth manifold?

 locally looks like Euclidian space  collection of charts

ll ibl

CS 176 Winter 2011

2

 mutually compatible

  • n their overlaps

 form an atlas

Parameterizations are key

Parameterizations

What is a parameterization?

 function from some region  ⊂ R2

to the embedded surface M⊂ R3

CS 176 Winter 2011

3

Image from Sander et al. 2002

Task

Find map from mesh to R2

 identify region on mesh and find

map to R2

n et al. 2002

CS 176 Winter 2011

4

 what’s different?

 existence of solutions when mapping

to convex region

Image from Desbrun

Applications

 texture mapping

 piecewise linear

interpolation of

Parameterizations

CS 176 Winter 2011

5

interpolation of attributes across mesh

 resampling  simulation

 needs surface as function

First Attempt

Tutte, 1963

 make planar vertex the barycenter

  • f its neighbors

 fix boundary

CS 176 Winter 2011

6

 fix boundary

 convex!

  • riginal

centroid mapping

slide-2
SLIDE 2

Conventions

How do we denote things?

 we are looking for 2 scalar valued

functions defined over the surface

CS 176 Winter 2011

7

 we will quietly ignore the

dimensionality of u and x vectors

 though it will be obvious from

context

Computing it

Solution of linear system

 input:  boundary: convex K-gon

d h

CS 176 Winter 2011

8

 edge weights:

Existence

Solvable?

 must show:

 i.e., no redundancy in equations

h t l ?

CS 176 Winter 2011

9

 what solver?

 system is large, (non-)symmetric  iterative

 Jacobi, Gauss-Seidel  CG, bi-CG

Setting the Weights

How to choose “barycenter”

 Tutte, “uniform”

 minimizes square lengths (springs)

CS 176 Winter 2011

10

Setting the Weights

How to choose the “barycenter”

 Floater, “shape preserving”

R3 R2

CS 176 Winter 2011

11

 geodesic polar map

Scaled angles Lengths

Setting the Weights

Floater weights

 barycentric coords??

CS 176 Winter 2011

12

slide-3
SLIDE 3

How well does it Work?

  • riginal salt dome

sample dataset

uniform inverse distance weighted shape preserving

CS 176 Winter 2011

13

sample approximations based on parameterization

all images from Floater, 1997

Smoothing Connection

Denoising surfaces

 move towards centroid in R3  connection with Laplacian

smoothing

CS 176 Winter 2011

14

smoothing

 Taubin, 1995  Desbrun, 1999  Guskov, 1999

Laplace Equation

Why is this connection useful?

 measures smoothness

 second derivatives (gradient squared)

“ti ” i i i it

CS 176 Winter 2011

15

 over “time”, minimizes it

Question

 how to discretize Laplace

Laplace Discretization

Taubin, 1995

 uniformity assumption

CS 176 Winter 2011

16

 smoothing of both

 geometry  parameterization

Umbrella Operator

Laplace Discretization

With collaborators, in 1999

 Laplace-Beltrami

 mean curvature flow

t k h i t t

CS 176 Winter 2011

17

 take shape into account

Mean Curvature Flow

With collaborators, in 1999

 also used implicit time stepping for

stability

CS 176 Winter 2011

18

  • riginal

Umbrella flow Mean curvature flow

slide-4
SLIDE 4

Measuring Distortion

Energy of a map

 discrete harmonic

 Eck, Polthier, Desbrun

CS 176 Winter 2011

19

domain range

Harmonic Map

Properties

 area minimization leads to minimal

surfaces: soap bubbles

CS 176 Winter 2011

20

Good Measures

Comparison of methods

 from Desbrun et al., 2002

 sensitivity to mesh shape

CS 176 Winter 2011

21

Kent et al., 1992 Floater, 1997 Sander et al., 2001 DAP DCP

Images from Desbrun et al., 2002

Variational Approach

Find parameterization which minimizes discrete energies

 DHP (harmonic)

CS 176 Winter 2011

22

 symmetric linear system… (nice)  need to fix boundary still

DHP

Properties

 used to build discrete conformal map  note that coefficients can be negative

(bad!)

CS 176 Winter 2011

23

(bad!)

 keeps angles but can result in very

large area distortion

Boundaries

How to fix them?

 Dirichlet boundary

 interpolation

k i l ith l ti

CS 176 Winter 2011

24

 e.g., k-gon on circle with relative

boundary edge length preserved

 “unnatural”

 Neumann boundary

 match derivatives on boundary

slide-5
SLIDE 5

Natural Boundary

Area gradient

 match on

boundary

Neumann Dirichlet

CS 176 Winter 2011

25 Neumann Dirichlet

i j k

Computational Issues

Computing charts

 face clustering

 all manner of issues…

b d diti

CS 176 Winter 2011

26

 boundary conditions

 map to particular boundary? which??

Linear system solvers

 iterative for large, sparse systems!

Recent Results

More formal def. of conformality

 see Schröder et al.

Non linearity creeps in

CS 176 Winter 2011

27

 but much stronger properties!