controlling anisotropy in mass-spring systems

19
iMAGIS is a joint project of CNRS - INPG - INRIA - UJF iMAGIS-GRAVIR / IMAG Controlling Anisotropy in Mass-Spring Systems David Bourguignon and Marie-Paule Cani iMAGIS-GRAVIR

Upload: david-bourguignon

Post on 13-Jul-2015

1.020 views

Category:

Technology


0 download

TRANSCRIPT

iMAGIS is a joint project of CNRS - INPG - INRIA - UJF

iMAGIS-GRAVIR / IMAG

Controlling Anisotropyin Mass-Spring Systems

David Bourguignon and Marie-Paule CaniiMAGIS-GRAVIR

iMAGIS-GRAVIR / IMAG

Motivation

•Simulating biological materials– elastic– anisotropic– constant volume deformation

•Efficient model⇒ mass-spring systems(widely used)

A human liver with themain venous systemsuperimposed

iMAGIS-GRAVIR / IMAG

Mass-Spring Systems

•Mesh geometry influences material behavior– homogeneity– isotropy

iMAGIS-GRAVIR / IMAG

Mass-Spring Systems

•Previous solutions– homogeneity

⇒ Voronoi regions [Deussen et al., 1995]

– isotropy/anisotropy⇒ parameter identification:simulated annealing, genetic algorithm[Deussen et al., 1995; Louchet et al., 1995]

⇒ hand-made mesh[Miller, 1988; Ng and Fiume, 1997]

Voronoi regions

v3

v2

v1

iMAGIS-GRAVIR / IMAG

Mass-Spring Systems

•No volume preservation

⇒ correction methods [Lee et al., 1995; Promayon et al., 1996]

iMAGIS-GRAVIR / IMAG

New Deformable Model

•Controlled isotropy/anisotropy⇒ uncoupling springs and mesh geometry

•Volume preservation

•Easy to code, efficient⇒ related to mass-spring systems

iMAGIS-GRAVIR / IMAG

Elastic Volume Element• Mechanical characteristics defined along axes of interest• Forces resulting from local frame deformation• Forces applied to masses (vertices)

I1’

I1 e1

e3

I3

I3’

e2

I2

I2’

I1’

I1 e1

A B

C

α

β

γ Barycenter Intersection points

iMAGIS-GRAVIR / IMAG

Forces Calculations

f1

I1’

I1

e1

f1’

f3

I1’

I1 e1

e3

I3

I3’f1

f1’

f3’

Stretch:Axial damped spring forces (each axis)

Shear:Angular spring forces(each pair of axes)

iMAGIS-GRAVIR / IMAG

F’1

Animation Algorithm

FC = γ F1 + γ’ F’1 + ...

FC

Example taken for atetrahedral mesh:

4 point masses3 orthogonal axes of interest

F1

I1’

I1 e1

2. Determine local frame deformation3. Evaluate resulting forces4. Interpolate to get resulting forces on vertices

xI = α xA + β xB + γ xC

A B

C

α

β

γ

I

1. Interpolate to get intersection points

iMAGIS-GRAVIR / IMAG

Animation Algorithm

xI = ζη xA + (1 – ζ)η xB +

(1 – ζ)(1 – η) xC + ζ(1 – η) xD

ζ

η

A B

CD

I

Interpolation scheme for anhexahedral mesh:

8 point masses3 orthogonal axes of interest

iMAGIS-GRAVIR / IMAG

Volume preservation• Extra radial forces• Tetra mesh: preserve sum of the barycenter-vertex distances• Hexa mesh: preserve each barycenter-vertex distance

With volume forces

Mass-spring system

Without volume forces

Tetrahedral Mesh

iMAGIS-GRAVIR / IMAG

Results

•Comparison with mass-spring systems:– no more undesired anisotropy– correct behavior in bending

Orthotropic material, same parameters in the 3 directions

iMAGIS-GRAVIR / IMAG

Results

•Control of anisotropy⇒ same tetrahedral mesh⇒ different anisotropic behaviors

iMAGIS-GRAVIR / IMAG

Results

Horizontal Diagonal Hemicircular

iMAGIS-GRAVIR / IMAG

Results

Concentric Helicoidal(top view)

RandomConcentric Helicoidal

iMAGIS-GRAVIR / IMAG

Results

•Performance issues: benchmarks on an SGI O2 (MIPS R5000 CPU 300 MHz, 512 Mb main memory)

Mesh Elements Springs / Element Time (in s)

Mass-Spring System Tetra 804 1.461 0.129

Hexa 125 8.320 0.117

Our Model Tetra 804 ≈10 1.867

Hexa 125 14 0.427

iMAGIS-GRAVIR / IMAG

Conclusion and Future Work

•Same mesh, different behaviors⇒ but different meshes, not the same behavior !

•Soft constraint for volume preservation

•Combination of different volume element types with different orders of interpolation

iMAGIS-GRAVIR / IMAG

Conclusion and Future Work

•Extension to active materials⇒ human heart motion simulation⇒ non-linear springs with time-varying properties

Angular maps of themuscle fiber direction in ahuman heart

iMAGIS-GRAVIR / IMAG