Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
130 views

Introduction To ParaView

ParaView is an open-source application for scientific data visualization and analysis. It has a graphical user interface that allows users to load data, apply filters to process and analyze the data, adjust display properties, and export images and data. ParaView uses a client-server architecture that enables parallel processing and visualization of large datasets. The document provides an overview of ParaView's capabilities and user interface, and includes examples of loading and visualizing a sample dataset to demonstrate common filters and workflows.

Uploaded by

juanarcos_778612
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
130 views

Introduction To ParaView

ParaView is an open-source application for scientific data visualization and analysis. It has a graphical user interface that allows users to load data, apply filters to process and analyze the data, adjust display properties, and export images and data. ParaView uses a client-server architecture that enables parallel processing and visualization of large datasets. The document provides an overview of ParaView's capabilities and user interface, and includes examples of loading and visualizing a sample dataset to demonstrate common filters and workflows.

Uploaded by

juanarcos_778612
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 41

Introduction to ParaView

Dan Lipsa,
R&D Engineer, Kitware Inc.

Contents
ParaView description, architecture and history
User interface: the Pipeline Browser and the Object
Inspector. ParaView objects: Filters, Representations and
Views. Finding Data and Selection.
Running ParaView in parallel.
Hands-on practice: vector visualization, data analysis

What is ParaView?
An open-source application and architecture for display and
analysis of scientific datasets.
Application - you dont have to write any code to analyze your data
Architecture - designed to be extensible if you want to code
Custom apps, plugins, python scripting, Catalyst, ParaViewWeb

Open-source BSD 3-clause license


Display - excels at traditional scientific vis qualitative 3D rendering
Analysis - data drill down through charts, stats, all the way to values
ParaView parallel - scales from notebooks to worlds largest
supercomputers

ParaView Architecture

Network

History
http://www.paraview.org/Wiki/ParaView_Release_Notes

1999 LANL/Kitware project (via ASCI Views)


Build an end user tool from VTK (TCL-TK interface)
Make VTK scale
October 2002 first public release, version 0.6

2002-2005 Versions 0.6 through 2.6


Continued growth under DOE Tri Labs, Army Research
Lab and various other partnerships

September 2005 ParaQ project started

Sandia, Kitware and CSimSoft


Make ParaView easier to use (Qt interface)
Add quantitative analysis
May 2007 version 3.0 released

Continuing to evolve
3.2, 3.4, 3.6, 3.8, 3.10, 3.12, 3.14, 3.98
4.0.1, 4.1, 4.2, 4.3.1 (January 2015)

User Interface
Menu Bar
Toolbars
Pipeline Browser
Object Inspector
Properties
Properties
Display
View
Information

Online help
6

View(s)

Help
Online help

The ParaView Guide (262 pages)


The ParaView Tutorial
ParaView Mailing Lists
ParaView Wiki
http://www.paraview.org/documentation/

How to Use ParaView

file

1. Read in data: File Open, hit Apply


reader

Over 100 file formats supported


Help/Readers - readers compiled in

2. Add a filter to process data:


Tune filter properties, hit Apply
Repeat Step 2 as needed

Edit Undo

3. Tune Display (for all Filter,View pairs) and


View (for all Views) parameters
4. Save datasets, rendered results
(screenshot or animation) or application
state

Filter = an object that operates on data: reads


its inputs and produces one or more outputs
Reader = reads a file and produces an output
Source = produces an output (Cylinder)
Pipeline object = a filter, reader or source

slice

warp

display

Filter Properties and the Apply Button


Active Filter properties
ParaView is meant to process large data it might take a long time
when changing a filter property.

Toggle auto
apply

ParaView Dataset Types


vtkImageData

vtkRectilinearGrid

vtkPolyData

vtkUnstructuredGrid

vtkStructuredGrid

Multi-blocks
AMR
Time-varying
data

10

- points, cells
- values associated points and/or cells: scalars,
vectors, tensors

Object Inspector
Information Tab
Information about the Active
Filters output
Dataset Type
Size (Bytes, #points, #cells)
Geometric bounds
Structured bounds
Arrays:

Name
Association =point,
Data Type
Data Ranges (and
scalar/vector)

Temporal Domain
11

=cell)

Multiple Views

12

Split (Vertical, Horizontal), Maximize, Close buttons


Active view: Display and View properties pertain to it
Right-click to link cameras

Display Properties
A Representation (a display): object that
stores visual characteristics of one
particular data set in one particular view
Properties associated with the Active Filter
and Active View.

13

Points

Wireframe

Surface

Surface
with Edges

Volume

Color Map Editor

Mapping
Scalar Range Color Palette
Rescale to data
range
Rescale to custom
range
Rescale to data
range over all
time-steps
Invert the
transfer function
Choose preset
Save to preset

14

Color Map (Transfer Function) Editor

15

View Properties
Properties associated with
the Active View

16

Find properties (for Filters, Displays and Views)

Search for properties


Toggle on/off advanced
properties
Advanced
Properties

17

Multi-View Visualization Pipeline


Representation

Source
Representation

Data
Representation

Filter

Representation

Data

Representation
Representation

Filter

View

Data
18

View

Pipeline Browser
Representation

Source
Representation

Data
Representation

Filter

Representation

Data

Representation
Representation

Filter

View

Data
19

View

Filter Properties acts on active filter


Representation

Source
Representation

Data
Representation

Filter

Representation

Data

Representation
Representation

Filter

View

Data
20

View

Information Tab shows output data for


the active filter
Representation

Source
Representation

Data
Representation

Filter

Representation

Data

Representation
Representation

Filter

View

Data
21

View

Display Properties representation


properties for active filter and view
Representation

Source
Representation

Data
Representation

Filter

Representation

Data

Representation
Representation

Filter

View

Data
22

View

View Properties acts on active view


Representation

Source
Representation

Data
Representation

Filter

Representation

Data

Representation
Representation

Filter

View

Data
23

View

Filters
Filters Menu

Recent
Common
Data Analysis
Statistical
Temporal
Alphabetical

Quick Launch
PC/Linux
CTRL-Space
Mac
ALT-Space

Apply Undo/Redo

24

Calculator

Glyph

Contour

Stream Tracer

Clip

Warp By Vector

Slice

Group Datasets

Threshold

Extract Group

Extract
Subset

Query Data by Attributes Values Find Data


Dialog

25

Query Data Visually - Selection


Visually select interesting data
shown in all compatible views
can then label, extract etc
Select Cells On to get nearest
cells
Select Points On
to get nearest points
Select Cells Through to get all
cells intersecting a frustum
Select Points Through
for selecting points inside a
frustum

26

Exporting data, images, sessions


File Save *
Active filters data, prompted for file format
List of file formats given in help primarily kitware
formats + exodus, ensight, xdmf/hdf5, csv

Screen shot, either selected view or all


Export visible scene in a format for high quality rendering
eps, pdf, ps, svg, pov, vrml, webgl, x3d, x3db

Movie
Image sequence, avi, ogg, ffmpeg avi

State
for restoring ParaView session later

27

What to Expect from Parallel Processing


Serial + Parallel = 1 (percentage that you cannot/can parallelize)

Amdahls Law
1
Speedup(CPUs)
Parallel
Serial
CPUs

Gustafsons Law
Speedup(CPUs) = CPUs *
Parallel + Serial

aka Strong scaling


If data size is fixed, dont
expect great scalability.
More processors != faster
aka Weak scaling
As data size grows, you
must have more resources.
More disk and IO = higher
resolution possible

What to Expect from Parallel


Processing

29

Parallel != Faster
Parallel Larger Problem Size

Large Data processed by ParaView

billion cell
weather
simulation

1 billion cell asteroid


detonation simulation

source: Sandia National Lab

30

Data Server
Data Server
Data Server
Data Server
Data Server
Data Server

Render Server
Render Server
Render Server
Render Server

Depth Composite

Client

Control,
Display and Rendering
of Small Data

N component Data
Parallelism for X GByte

Reader
X/N GB

31

MPI

Reader
X/N GB

Filter

Filter

Filter

Filter

Tile Display

ParaViews Running Modes


Builtin aka
Standalone
aka Serial

Combined
Server

Batch

DS

RS

RS
DS
RS
DS
RS
DS

RS
DS
RS
DS
RS
DS

DS = data server
RS = render server
32

Client

all components within one process


(client may be GUI or pvpython)
paraview || pvpython

data processing and parallel rendering in


MPI job of combined processes. control
Client from TCP connected client.
mpiexec -n x pvserver &;
paraview # or pvpython #+ Connect

server is an MPI job which directly


runs a python script
mpiexec n x pvbatch \
vis_script.py

Connecting to a Server
Follow instructions at http://www.alcf.anl.gov/user-guides/paraview-cooley
File Connect

33

Level of Detail Maintain Interactivity


Type 1: Geometrically based
Edit Settings Render View
LOD threshold = 0.1
Down-samples geometry while
interacting

34

Level of Detail Maintain Interactivity


Type 2: Image Based

35

Edit Settings Render View


Remote Render Threshold = 0.1
Image Reduction Factor = 10
Down-samples pixels while interacting

Hands on practice: vector visualization


(see also http://www.paraview.org/Wiki/The_ParaView_Tutorial)

file

Load disk_out_ref.ex2
An exodus format file
Enable all variables

representation

Information tab

Multi-block (group of datasets)


Not time varying
Roughly 8000 cells and points, 2MB
11.5 units in diameter, 20 units in
height

Show as surface with edges to see


structure
Set opacity to 0.5
Looks like a cylinder with a recess
view 1
36

reader

Hands on practice: vector visualization

file

Apply slice filter


Align with z and use
10 offset values

Color by Temp
Show temp color key
Adjust opacity of reader(0.1) and
slice(1.0) so that you can see
temperature variation clearly

37

representation

reader

representation

slice

Hands on practice: vector visualization

file

Apply warp filter


Warp slices along V vector field
with a scale factor of 0.1

reader

Compare with display of slice


note see how vector field
pushes up in center and down
further out
We are seeing convection
of a heated gas, it rises at
the heat source

38

slice

warp

Hands on practice: vector visualization

file

Change warp opacity to .2


Apply streamline filter
Starts from seed points and advects along
vector field to show you vector flow

reader

Apply tube filter


Gives infinitely thing streamlines extent so
we can see them well

stream
line

slice

tube

warp

Set opacity to 1.0 and


color by vorticity
We are seeing rotation
A heated plate is
spinning in gas

Manipulate streamlines
seed points

39

Putting it together: data analysis


reader
view 2

view 3

view 4
40

plot over
line

histogram

Thank You.
Questions?

41

You might also like