[email protected] 1 Feature Extraction and Visualization of Flow Fields State-of-the-Art Report Feature Extraction and Visualization of Flow Fields Frits.

1Feature Extraction and Visualization of

Flow [email protected]

State-of-the-Art Report

Feature Extraction and Visualization of Flow

FieldsFrits H. Post and Benjamin Vrolijk

Delft University of TechnologyThe Netherlands

visualisation.tudelft.nl{F.H.Post,B.Vrolijk}@its.tudelft.nl

Helwig Hauser, Robert S. Laramee, Helmut Doleisch

VRVis Research CenterAustria

www.VRVis.at{hauser,laramee,doleisch}@VRVis.at

Part 2:Feature-Based FlowViz

Part 1: Visualization of Flow Fields



Flow Visualization STAR Overview

Part 1: Visualization of Flow Fields

Introduction to Flow Visualization (FlowViz)

Direct FlowViz Texture-Based FlowViz Geometric FlowViz

Part 2: Feature-Based FlowViz



The FlowViz Job

Goal: communicating FlowViz data: data representing fluid/gas flow, i.e.

vector field data (magnitude + direction) not just scalar data visualization is a very high bandwidth

channel

User Goals: obtain overview of vector field present characteristics identify and investigate details and

features



Computational vs. Experimental and Empirical FlowVis

Computational FlowVis -using computers for FlowVis data resulting from flow simulation,

measurements, or flow modelling, e.g., computational fluid dynamics (CFD)

computer-generated images and animations, often mimicking experimental FlowVis

Visualization of actual fluids, e.g. water and air dye injection interferometry Schlieren/shadows flow topology graphs etc.



FlowVis Data vs. Data AcquisitionData from Simulation: FOR EACH cell in (irregular) grid: compute flow direction compute flow magnitude (explicitly or implicitly) compute pressure (à la Navier-Stokes equations) further attributes

Data from measurements: FOR EACH location of a (regular) grid: flow direction (reconstructed) measure flow magnitude

Data from modelling: Vector Field represented by analytic

function Flow direction and magnitude a

function of location (and time)



FlowViz Data Characterized by Many Dimensions

Spatial dimensions: 2D (planar flow, simplified or synthetic) 2.5D (boundary flow, flow on surface) 3D (real-world flow)Temporal dimension: steady flow -1 time step (or instantaneous flow) time-dependent flow -multiple time steps (turbulent,

real-world) caution is advised in the context of animationData dimensions: velocity temperature pressure and many more...



Direct vs. Geometric vs. Feature-Based FlowViz

focus on featuresmore detailed viewOverview, 1st impression



FlowViz Fundamentals

Flow Data from simulation: vector field represented as samples: vp,t (+ reconstruction)

v(p,t) = dp/dt; p,v Rn, t R

Flow Integration over Grids: Cartesian, curvi-linear, unstructured integration techniques: Euler and Runge-Kutta point location (which cell p lies in) and neighbor

searching flow reconstruction within a cell (à la

interpolation) computation of derived data

p(s) = p0 + v(p(), +t0) d [instantaneous]

pE(t+t) = p(t) + t v(p(t),t) [time-dependent]



Direct Flow VisualizationDirect Mapping of Flow Attributes to Visualization

Space

Advantages: simplicity less computation time intuitive

Disadvantages: does not always clearly

show flow properties and features, e.g., flow orientation



Direct FlowViz: Color Coded Slicing

Color Coding in 2D, instantaneous: mapping flow attribute(s) to hue

slicing probe for vortex visualization

(Schulz et al)

multiple slices for vortex visualization (missing color coded boundary)



Direct FlowViz: Arrow/Hedgehog plots in 2D and 3D, instantaneous



Direct FlowViz: Contours in 2D and isosurfaces in 3D

Contours in a slice

Isosurface (and color coding) in 3D



Direct FlowViz: Volume Rendering combined with color coding

the curvilinear bluntfin data set(Westermann)



Direct FlowViz: Hybrid Solutions -arrow plots and color coding in 2D, steady and

unsteady



Texture-Based Flow Visualization

Computing textures that provide a dense coverage/visualization of a vector field

Advantages: detailed view of vector field clearer perception of

characteristics contains elements of direct

+ geometric FlowViz

Disadvantages: computation

time 2.5D, 3D



Texture-Based FlowViz: Spot Noise (Van Wijk) and LIC (Cabral and Leedom) in 2D,

instantaneous



Texture-Based FlowViz: Spot Noise in 2D, instantaneous, with color coding (de Leeuw)

Good for visualizing detail.



Texture-Based FlowViz: Spot Noise and LIC in 2D, time-dependent (Van Wijk)



Texture-Based FlowViz: Time-Dependent Texture Advection in 2D

Unsteady FlowViz of the Gulf of Mexico (Jobard et al)



Texture-Based FlowViz: LIC on Surfaces, unsteady

A comparison of 3 LIC techniques (left) UFLIC, (middle) ELIC, and (right) PLIC (Verma et. al.)



Texture-Based FlowViz: LIC in 3D, instantaneous (Interrante and Grosch)



Geometric Flow Visualization

The computation of objects whose shape is directly related to underlying geometry

Advantages: intuitive clearer perception of

characteristics

Disadvantages: placement 3D



Geometric FlowViz: Some Terminology

Stream vs. Path vs Streak vs Time lines

Streamline everywhere tangent to flow at

instantaneous time, t0 (blue/aqua)

Pathline path traced by a particle

over time, t (red/maroon)

Streakline line traced by continuous

injection at location, x0 (light green)

Timeline temporal evolution of initial

line, l0 (yellow)



Geometric FlowViz: Streamlines and Streamlets in 2D, steady-state



Geometric FlowViz: Pathlines and Streamlets in 2D, unsteady (Van Wijk)



Geometric FlowViz: Timelines in 2D (Van Wijk) and 3D (B. Girod) (unsteady)



Geometric FlowViz: Seeding in 2D (B. Jobard) and 3D (Schulz et al)

Image-based, topology-based, and interactive seeding strategies



Geometric FlowViz: Streamribbons and Streamtubes, 3D, steady-state



Geometric FlowViz: Perceptual Issues in 3D, steady-state

Illuminated Streamlines (Zoeckler)

StreamRunner (Laramee)



Geometric FlowViz: Streaklines in 2D (Jobard et al) and 3D (B. Girod)



Geometric FlowViz: StreamBalls (Brill et al), StreamSurfaces (Hultquist),

StreamArrows (Loeffelmann et al), 3D, steady-state



Geometric FlowViz: Flow Volumes (Crawfis), steady and unsteady

A subset of 3D flow domain specified by initial 2D patch



Some Open Issues in FlowViz

Unsteady FlowViz on Surfaces -esp. Texture-based, unstructured

Steady-State FlowViz in 3D -perceptual issues, seeding strategies

Unsteady FlowViz in 3D -computation time

Lot’s of work to (still) be done.



Acknowledgements: Part 1

Thanks to (1) the KPlus (www.kplus.at) -Austrian governmental research program, and (3) AVL (www.avl.com) for financial support

For more information see: www.VRVis.at or email [email protected]

Now for Part 2!

[email protected] 1 Feature Extraction and Visualization of Flow Fields State-of-the-Art Report Feature Extraction and Visualization of Flow Fields Frits.

Documents