Start ParaView on OnDemand

Status: draft. Verified steps and screenshots to come.

In this lesson you will open a dataset in ParaView on a compute node, color it by a variable, and save a picture. Time: 10 minutes · Branch: Your program · Program: ParaView 6 (AMA27 ParaView module)

Before this lesson
Before this Data You'll need
Start an OnDemand session disk_out_ref.ex2: air flowing around a heated, spinning disk. It comes with ParaView (BSD-3) Nothing else

Note

Cluster details. Highlighted values are placeholders. AMA27's queue names, module versions, paths and addresses go in when the cluster is deployed.

Why this matters

ParaView reads the formats simulation codes usually write: Exodus, VTK, NetCDF, OpenFOAM, CSV and image stacks. It can work on datasets larger than a laptop's memory. Running it on the cluster leaves the data where it was written. The compute node reads it and draws the picture, and only the picture travels to your browser.

Key idea

The pipeline. ParaView builds a picture as a chain: a reader loads the data, filters change it, and each item can be shown or hidden. The Pipeline Browser on the left is that chain. Nothing in it changes until you press Apply. Each step reads its input and writes a new output. The file on disk is left alone, and any step can be adjusted later. → Glossary: visualization pipeline

Steps

Step 1: Start ParaView

On the OnDemand dashboard choose Interactive Apps ▸ ParaView. Fill in the form as in the start lesson: AMA27 queue name, 1 hour, 4 cores. Launch, then connect.

You should see: the ParaView window in your browser, with the Pipeline Browser on the left and an empty render view in the middle.

Screenshot: the Pipeline Browser and render view, marked

Screenshot to come: ParaView's window, with the Pipeline Browser on the left and the empty render view in the middle marked

Tip

No ParaView app in your list? Start the desktop app (AMA27 desktop app name) instead, open Applications ▸ Terminal Emulator, load ParaView (AMA27 ParaView module) and run paraview &.

Step 2: Open the example data

Choose File ▸ Open…, click Examples in the left-hand list, select disk_out_ref.ex2 and click OK. In the Properties panel, make sure every variable is ticked, then click Apply.

You should see: a gray, roughly cylindrical shape in the render view, and disk_out_ref.ex2 in the Pipeline Browser.

Screenshot: the Open File dialog, with Examples marked

Screenshot to come: ParaView's Open File dialog, with Examples marked in the left-hand list and disk_out_ref.ex2 selected

Note

No Examples entry? Download the same files into scratch with Step 3 of Get your data where the app can see it, then open ~/scratch/ParaViewTutorialData-20220629/disk_out_ref.ex2.

Step 3: Move around

In the render view:

  • drag with the left mouse button to rotate;
  • scroll, or drag with the right button, to zoom;
  • drag with the middle button (or Shift + left) to move sideways.

The Reset Camera button on the camera toolbar brings the whole object back into view.

You should see: the shape turn and move as you drag.

Screenshot: the Reset Camera button

Screenshot to come: The camera toolbar above the render view, with the Reset Camera button marked

Step 4: Color by temperature

With disk_out_ref.ex2 selected, open the coloring drop-down on the toolbar (it says Solid Color) and choose Temp. Then click Rescale to Data Range on the same toolbar.

You should see: the surface colored from cool to hot, with a color legend labeled Temp.

Screenshot: the coloring drop-down and the legend

Screenshot to come: The toolbar's coloring drop-down set to Temp, and the color legend labeled Temp in the render view

Note

Some ParaView versions show the variable names in lower case (temp, pres, v). Pick whichever your list shows.

Step 5: Use a color map that doesn't mislead

In the Properties panel, find Coloring and click Choose Preset (the icon beside Edit). Search for Viridis, choose it, click Apply, then Close.

You should see: colors running from dark purple (cool) to yellow (hot). Viridis keeps the order of its colors in grayscale and for red-green color blindness, which a rainbow color map does not.

What you should see

Picture to come: The practice data as a roughly cylindrical shape colored by temperature with Viridis: dark purple where cool, yellow where hot

Screenshot: the Choose Preset dialog

Screenshot to come: The Choose Preset dialog, with Viridis typed in the search box and selected

Step 6: Save a picture

Choose File ▸ Save Screenshot…. Set the file name to ~/scratch/paraview-first.png and the size to 1600 by 1200, then click OK.

You should see: paraview-first.png in the Files app, in your scratch folder.

Screenshot: the Save Screenshot dialog

Screenshot to come: The Save Screenshot dialog, with the image size set to 1600 by 1200

Step 7: Finish

Choose File ▸ Exit to close ParaView. Then delete the session on My Interactive Sessions, as in the start lesson.

You should see: the session card gone.

Check your understanding

After choosing a new variable, the whole object is one flat color. What single click usually fixes it?

Rescale to Data Range. The color scale was still set for the previous variable's values, so every value in the new one fell at one end of it.

Where did ParaView actually run, and why does that matter for a 200 GB dataset?

On one of the cluster's compute nodes. The data is read where it already lives, and only the rendered picture is sent to your browser. The size of the file has little to do with your connection.

If something goes wrong

If something goes wrong
What you see Why What to do
Apply stays highlighted and nothing appears The change hasn't been applied Click Apply
The render view stays black ParaView draws with the node's processors and is still working Give it a few seconds, or click inside the view
Rotating is slow and jerky The picture is large, or the image quality is set high Make the window smaller, or lower Image Quality on the session card
Temp isn't in the list Not all variables were loaded Select the reader, tick every variable in Properties, click Apply

The ParaView family

ParaView is several programs that share one engine. The lessons use:

The ParaView family
Tool What it's for Taught in
paraview This window: interactive exploration this lesson; the density and vector-field lessons
View ▸ Python Shell, Tools ▸ Start Trace Turning clicks into Python From clicks to a script
Python Calculator Computing new fields, NumPy-style (curl, gradient, mag) Derived fields: vorticity
Programmable Filter Any Python inside the pipeline Derived fields (going further)
pvbatch and pvpython Running ParaView Python scripts, with or without a window From clicks to a script, Batch rendering
pvserver The server half of client and server use module B5, "Where next"

Next

Credits

Data: disk_out_ref.ex2, ParaView example data (Kitware/Sandia), BSD-3-Clause. ParaView is BSD-3-Clause. Menu names follow the ParaView 6.0.1 documentation. Written for SBSC; text CC-BY 4.0.