4. Displaying data

The goal of any visualization process is to produce visual representations of the data. The visual representations are shown in modules called views. Views provide the canvas on which to display such visual representations, as well as to dictate how these representations are generated from the raw data. The role of the visualization pipeline is often to transform the data so that relevant information can be represented in these views.

Referring back to the visualization pipeline from Section 1.2, views are sinks that take in input data but do not produce any data output (i.e., one cannot connect other pipeline modules such as filters to process the results in a view). However, views often provide mechanisms to save the results as images or in other formats including PDF, VRML, and X3D.

Different types of views provide different ways of visualizing data. These can be broadly grouped as follows:

• Rendering Views are views that render geometries or volumes in a graphical context. The Render View is one such view. Other Render View-based views, such as Slice View and Quad View , extend the basic render view to add the ability to add mechanisms to easily inspect slices or generate orthogonal views.

• Chart Views cover a wide array of graphs and plots used for visualizing non-geometric data. These include views such as line charts ( Line Chart View ), bar charts ( Bar Chart View ), bag charts ( Bag Chart View ), parallel coordinates ( Parallel Coordinates View ), etc.

• Comparative Views are used to quickly generate side-by-side views for parameter study, i.e., to visualize the effects of parameter changes. Comparative variants of Render View and several types of the Chart Views are available in ParaView.

In this chapter, we take a close look at the various views available in ParaView and how to use these views for displaying data.

4.1. Multiple views

Fig. 4.1 Using multiple views in paraview to generate different types of visualizations from a dataset.

With multiple types of views comes the need for creating and viewing multiple views at the same time. In this section, we look at how you can create multiple views and lay them out.

Did you know?

Multiple views were first supported in ParaView 3.0. Before that, all data was shown in a single 3D render view, including line plots!

4.1.1. Multiple views in paraview

paraview shows all views in the central part of the application window. When paraview starts up, the Render View is created and shown in the application window by default.

New views can be created by splitting the view frame using the Split View controls at the top-right corner of the view frame. Splitting a view divides the view into two equal parts, either vertically or horizontally, based on the button used for the split. On splitting a view, an empty frame with buttons for all known types of views is shown. Simply click on one of those buttons to create a new view of a chosen type.

You can move views by clicking and dragging the title bar for the view (or empty view frame) and dropping it on the title bar on another view (or empty view frame). This will swap the positions of the two views.

Similar to the notion of active source, there is a notion of active view . Several panels, toolbars, and menus will update based on the active view. The Display properties section on the Properties panel, for example, reflects the display properties of the active source in the active view. Similarly, the eyeball icons in the Pipeline Browser show the visibility status of the pipeline module in the active view. Active view is marked in the UI by a blue border around the view frame. Only one view can be active at any time in the application.

Besides being able to create multiple views and laying them out in a pane, paraview also supports placing views in multiple layouts under separate tabs. To create new tabs, use the button in the tab bar. You can close a tab, which will destroy all views laid out in that tab, by clicking on the button. To pop out an entire tab as a separate window, use the button on the tab bar.

The active view is always present in the active tab. Thus, if you change the active tab, the active view will also be changed to be a view in the active tab layout. Conversely, if the active view is changed (by using the Python Shell , for example), the active tab will automatically be updated to be the tab that contains the active view.

Did you know?

You can make the views of the active layout fullscreen by using View > Fullscreen (layout) (or using the F11 key). You can also make the active view alone fullscreen by using View > Fullscreen (active view) (or using CTRL + F11 keys). To return back to the normal mode, use the Esc key.

4.1.2. Multiple views in pvpython

In pvpython, one can create new views using the CreateView function or its variants, e.g., CreateRenderView .

>>> from paraview.simple import *
>>> view = CreateRenderView()
# Alternatively, use CreateView.
>>> view = CreateView("RenderView")

When a new view is created, it is automatically made active. You can manually make a view active by using the SetActiveView function. Several of the functions available in pvpython will use the active view when no view is passed as an argument to the function.

# Create a view
>>> view1 = CreateRenderView()
# Create a second view
>>> view2 = CreateRenderView()

# Check if view2 is the active view
>>> view2 == GetActiveView()
True

# Make view1 active
>>> SetActiveView(view1)
>>> view1 == GetActiveView()
True

When using Python Shell in paraview, if you create a new view, it will automatically be placed in the active tab by splitting the active view. You can manually control the layout and placement of views from Python too, using the layout API.

In Python, each tab corresponds to a layout.

# To get exisiting tabs/layouts
>>> layouts = GetLayouts()
>>> print(layouts)
{('ViewLayout1', '264'): <paraview.servermanager.ViewLayout object at 0x2e5b7d0>}

# To get layout corresponding to a particular view
>>> print(GetLayout(view))
<paraview.servermanager.ViewLayout object at 0x2e5b7d0>

# If view is not specified, active view is used
>>> print(GetLayout())
<paraview.servermanager.ViewLayout object at 0x2e5b7d0>

# To create a new tab
>>> new_layout = servermanager.misc.ViewLayout(registrationGroup="layouts")

# To split the cell containing the view, either horizontally or vertically
>>> view = GetActiveView()
>>> layout = GetLayout(view)
# fraction is optional, if not specified the frame is split evenly.
>>> locationId = layout.SplitViewVertical(view=view,
                                          fraction=0.5)
# To get the location of the layout
>>> locationId = layout.GetViewLocation(view)

# To change the split factor
>>> layout.SetSplitFraction(locationId, 0.75)

# To Maximize a particular view
>>> layout.MaximizeCell(locationId)

# To assign a view to a particular cell.
>>> view2 = CreateRenderView()
>>> layout.AssignView(locationId, view2)

4.2. View properties

Just like parameters on pipeline modules, such as readers and filters, views provide parameters that can be used for customizing the visualization such as changing the background color for rendering views and adding title texts for chart views. These parameters are referred to as View Properties and are accessible from the Properties panel in paraview.

4.2.1. View properties in paraview

Similar to properties on pipeline modules like sources and readers, view properties are accessible from the Properties panel. These are grouped under the View section. When the active view is changed, the Properties panel updates to show the view properties for the active view. Unlike pipeline modules, however, when you change the view properties, they affect the visualization immediately, without use of the Apply button.

Did you know?

It may seem odd that View and Display properties on the Properties panel don’t need to be Apply -ed to take effect, while properties on pipeline modules like sources, readers and filter require you to hit the Apply button.

To understand the reasoning behind that, we need to understand why the Apply action is needed in the first place. Generally, executing a data processing filter or reader is time consuming on large datasets. If the pipeline module keeps on executing as you are changing the parameter, the user experience will quickly deteriorate, since the pipeline will keep on executing with intermediate (and potentially invalid) property values. To avoid this, we have the Apply action. This way, you can set up the pipeline properties to your liking and then trigger the potentially time consuming execution.

Since the visualization process in general focuses on reducing data to generate visual representations, the rendering (broadly speaking) is less time-intensive than the actual data processing. Thus, changing properties that affect rendering are not as compute-intensive as transforming the data itself. For example, changing the color on a surface mesh is not as expensive as generating the mesh in the first place. Hence, the need to Apply such properties becomes less relevant. At the same time, when changing display properties such as opacity, you may want to see the result as you change the property to decide on the final value. Hence, it is desirable to see the updates immediately.

Of course, you can always enable Auto Apply to have the same immediate update behavior for all properties on the Properties panel.

4.2.2. View properties in pvpython

In pvpython, once you have access to the view, you can directly change view properties on the view object. There are several ways to get access to the view object.

# 1. Save reference when a view is created
>>> view = CreateView("RenderView")

# 2. Get reference to the active view.
>>> view = GetActiveView()

The properties available on the view will change based on the type of the view. You can use the help function to discover available properties.

>>> view = CreateRenderView()
>>> help(view)

  Help on RenderView in module paraview.servermanager object:

class RenderView(Proxy)
 |  View proxy for a 3D interactive render
 |  view.
 |
 |  ----------------------------------------------------------------------
 |  Data descriptors defined here:
 |
 |  CenterAxesVisibility
 |      Toggle the visibility of the axes showing the center of
 |      rotation in the scene.
 |
 |  CenterOfRotation
 |      Center of rotation for the interactor.
 |
 ...

# Once you have a reference to the view, you can then get/set the properties.

# Get the current value
>>> print(view.CenterAxesVisibility)
1

# Change the value
>>> view.CenterAxesVisibility = 0

4.3. Display properties

Display properties refers to available parameters that control how data from a pipeline module is displayed in a view, e.g., choosing to view the output mesh as a wireframe, coloring the mesh using a data attribute, and selecting which attributes to plot in chart view. A set of display properties is associated with a particular pipeline module and view. Thus, if the data output from a source is shown in two views, there will be two sets of display properties used to control the appearance of the data in each of the two views.

4.3.1. Display properties in paraview

Display properties are accessible from the Display section on the Properties panel. When the active source or active view changes, this section updates to show the display properties for the active source in the active view, if available. If the active source produces data that cannot be shown (or has never been shown) in the view, then the Display properties section may be empty.

Similar to view properties, display property changes are immediately applied, without requiring the use of the Apply button.

4.3.2. Display properties in pvpython

To access display properties in pvpython, you can use SetDisplayProperties and GetDisplayProperty methods.

# Using SetDisplayProperties/GetDisplayProperties to access the display
# properties for the active source in the active view.

>>> print(GetDisplayProperties("Opacity"))
1.0

>>> SetDisplayProperties(Opacity=0.5)

Alternatively, you can get access to the display properties object using GetDisplayProperties and then changing properties directly on the object.

# Get display properties object for the active source in the active view.
>>> disp = GetDisplayProperties()

# Get the available representation types. Available is an option for all properties that have a Domain.
>>> disp.GetProperty("Representation").Available
['Outline', 'Points', 'Wireframe', 'Surface', 'Surface With Edges']

# You can also save the object returned by Show.
>>> disp = Show()

# Now, you can directly access the properties.
>>> print(disp.Opacity)
0.5

>>> disp.Opacity = 0.75

As always, you can use the help method to discover available properties on a display object.

>>> disp = Show()

>>> help(disp)
>>> help(a)
Help on GeometryRepresentation in module paraview.servermanager object:

class GeometryRepresentation(SourceProxy)
 |  ParaView`s default representation for showing any type of
 |  dataset in the render view.
 |
 |  Method resolution order:
 |      GeometryRepresentation
 |      SourceProxy
 |      Proxy
 |      __builtin__.object
 |
 |  ----------------------------------------------------------------------
 |  Data descriptors defined here:
 |
 |  ...
 |
 |  CenterStickyAxes
 |      Keep the sticky axes centered in the view window.
 |
 |  ColorArrayName
 |      Set the array name to color by. Set it to empty string
 |      to use solid color.
 |
 |  ColorAttributeType
 |  ...

4.4. Render View

The Render View is the most commonly used view in ParaView. It is used to render geometries and volumes in a 3D scene. This is the view that you typically think of when referring to 3D visualization. The view relies on techniques to map data to graphics primitives such as triangles, polygons, and voxels, and it renders them in a scene.

Most of the scientific datasets discussed in Section 3.1 are composed of meshes. These meshes can be mapped to graphics primitives using several of the established visualization techniques. That is, you can compute the outer surface of these meshes and then render that surface as filled polygons, you can just render the edges, or you can render the data as a nebulous blob to get a better understanding of the internal structure in the dataset. Plugins, like Digital Rock Physics, can provide additional ways of rendering data using advanced techniques that provide more insight into the data.

If the dataset doesn’t represent a mesh, e.g., a table (Section 3.1.9), you cannot directly show that data in this view. However, in such cases, it may be possible to construct a mesh by mapping columns in the table to positions to construct a point cloud, for example.

Fig. 4.2 paraview using Render View to generate 3D visualizations from a dataset.

4.4.1. Understanding the rendering process

Render View uses data processing techniques to map raw data to graphics primitives, which can then be rendered in a 3D scene. These mapping techniques can be classified as follows:

• Surface rendering methods provide general rendering by rendering a surface mesh for the dataset. For polygonal datasets (Section 3.1.8), this is simply the raw data. In cases of other datasets including structured (Section 3.1.3, Section 3.1.4, Section 3.1.5) and unstructured (Section 3.1.7) grids, this implies extracting a surface mesh for all external faces in the dataset and then rendering that mesh. The surface mesh itself can then be rendered as a filled surface or as a wireframe simply showing the edges, etc.

• Slice rendering is available for uniform rectilinear grid datasets (Section 3.1.3) where the visualization is generated by simply rendering an orthogonal slice through the dataset. The slice position and alignment can be selected using the display properties.

• Volume rendering generates rendering by tracing a ray through the dataset and accumulating intensities based on the color and opacity transfer functions set.

Each of these techniques are referred to as representations. When available, you can change the representation type from the display properties on the Properties panel or using the Representation Toolbar .

4.4.2. Render View in paraview

4.4.2.1. Creating a Render View

Unless you changed the default setting, a new Render View will be created when paraview starts up or connects to a new server. To create a Render View in paraview, split or close a view, and select the Render View button. You can also convert a view to a Render View (or any other type) by right-clicking on the view’s title bar and picking from the Convert To sub-menu. It simply closes the chosen view and creates a selected view type in its place.

You can use the Pipeline Browser to control the visibility of datasets produced by pipeline modules in this view. The eyeball icons reflect the visibility state. Clicking on the eyeball icon will toggle the visibility state. If no eyeball icon is shown, it implies that the pipeline module doesn’t produce a data type that can be directly shown in the active view, e.g., if the module produced a table, then when Render View is active, there will not be any eyeball icon next to that module.

4.4.2.2. Interactions

You can interact with the Render View to move the camera in the scene for exploring the visualization and setting up optimal viewing angles. Each of the three mouse buttons, combined with keyboard modifier keys (CTRL or ⌘ , and ⇧ ), move the camera differently. The interaction mode can be changed from the Camera tab in the Settings dialog, which is accessible from Tools > Settings (or ParaView > Preferences on macOS). There are six interaction modes available in ParaView:

• Pan for translating the camera in the view plane.

• Zoom for zooming in or out of the center of the view.

• Roll for rolling the camera.

• Rotate for rotating the camera around the center of rotation.

• Zoom To Mouse for zooming in or out of the projected point under the mouse position.

• Multi Rotate for allowing azimuth and elevation rotations by dragging from the middle of the view and rolls by dragging from the edges.

• Rotate Skybox for rotating the environment skybox. Useful when using Environment Lighting and PBR shader.

The default interactions options are as follows:

Modifier	Left Button	Middle Button	Right Button
	Rotate	Pan	Zoom
⇧	Roll	Rotate	Pan
CTRL or ⌘	Rotate Skybox	Rotate	Zoom To Mouse

Usually in ParaView, you are interacting with a 3D scene. However, there are cases when you are working with a 2D dataset such as a slice plane or a 2D image. In such cases, paraview provides a separate set of interaction options suitable for 2D interactions. You can toggle between the default 3D interaction options and 2D interaction options by clicking the 2D or 3D button in the view toolbar. The default interaction options for 2D interactions are as follows:

Modifier	Left Button	Middle Button	Right Button
	Pan	Roll	Zoom
⇧	Zoom	Zoom	Zoom To Mouse
CTRL or ⌘	Roll	Pan	Rotate

By default, ParaView will determine whether your data is 2D or 3D when loading the data and will set the interaction mode accordingly. This behavior can be changed in the Settings dialog by changing the Default Interaction Mode setting under the Render View tab. The default setting is “Automatic, based on the first time step”, but the setting can be changed to “Always 2D” or “Always 3D” in case you wish to force the interaction mode.

4.4.2.3. View properties

Several of the view properties in Render View control the annotations shown in the view (Fig. 4.3).

Fig. 4.3 The Properties panel showing view properties for Render View .

Axes Grid refers to an annotation axis rendered around all datasets in the view (Fig. 4.4). You use the checkbox next to the Edit Axis Grid button to show or hide this annotation. To control the annotation formatting, labels, etc., click on the Edit Axes Grid... button. The Axes Grid is described in Chapter Section 11.

Fig. 4.4 Axes Grid is used to annotate data bounds in Render View .

The Center axes refers to axes rendered in the scene positioned as the center of rotation, i.e., the location is space around which the camera revolves during Rotate camera interaction. Center Axes Visibility controls the visibility of the center axes.

The Orientation axes is the widget shown at the lower-left corner by default, which is used to get a sense for the orientation of the scene. The properties grouped under the Orientation Axes group allow you to toggle the visibility and the interactivity of this widget. When interactivity is enabled, you can click and drag this widget to the location of your choosing in the scene.

You can also change the Background used for this view. You can either set it as a Single color or as a Gradient changing between two colors, or you can select an Image (or texture) to use as the background.

There are two advanced properties you may wish to set: hidden line removal and camera parallel projection. The Hidden Line Removal option can be enabled to hide lines that would be occluded by a solid object when drawing objects in a Wireframe representation. If you want to render your data using parallel projection instead of the default perspective projection you can check the Camera Parallel Projection checkbox.

4.4.2.4. Display properties

One of the first (and probably the most often used) display properties is Representation . Representation allows you to pick one of the mapping modes. The options available depend on the data type, as well as the plugins loaded. While all display properties are accessible from the advanced view for the Properties panel, certain properties may be shown/hidden from the default view based on the chosen representation type.

Fig. 4.5 Different renderings generated by rendering data produced by the Wavelet source as outline, points, slice, surface, surface with edges, and volume.

The Outline representation can be used to render an outline for the dataset. This is arguably the fastest way of rendering the dataset since only the bounding box is rendered. Scalar coloring options, i.e., selecting an array with which to color, has no effect on this representation type. You can still, however, change the Solid Color to use as well as the Opacity . To change the color, select Solid Color in the combo-box under Coloring , and then click Edit to pick the color to use. To change the opacity, simply change the Opacity slider. 0 implies total transparency and, hence, invisiblity, while 1 implies totally opacity.

Did you know?

Rendering translucent data generally adds computational costs to the rendering process. Thus, when rendering large datasets, you may want to leave changing opacities to anything less than 1 to the very end, after having set up the visualization. In doing so, you avoid translucent geometries during exploration, but use them for generating images or screenshots for presentations and publications.

Points , Surface , Surface With Edges , and Wireframe rely on extracting the surface mesh from the dataset and then rendering that either as a collection of points, as solid surface, as solid surface with cell boundaries highlighted, or as a wireframe of cell boundaries only. Feature Edges is a subset of Wireframe consisting of prominent edges on the surface such as edges between cells that form a sharp angle or edges with only one adjacent cell. For these representations, you can either set a single solid color to use, as with Outline , or select a data array to use for scalar coloring (also known as pseudocoloring).

Two other representations are available for most datasets. 3D Glyphs draws a copy of a 3D geometry (e.g., arrow, cone, or sphere, etc.), or glyph, at a subset of points in the dataset. These glyphs can be set to a single color or pseudocolored by a data array. The Point Gaussian representation is similar, but instead of drawing 3D geometry at every point, it draws a 2D image sprite that may have transparency. The image drawn can be one of several predefined image sprites such as Gaussian Blur , Sphere , Black-edged circle , Plain cirlce , Triangle , or Square outline , or a custom sprite can be defined with custom GLSL shader code.

Lastly, the SurfaceLIC representation is available for surface datasets with vector point data arrays. LIC stands for line integral convolution, which is a visualization technique that shows the direction of flow as a noise pattern smeared in the direction of flow.

Fig. 4.6 An example of the SurfaceLIC representation showing the direction of a vector data array and colored by a different scalar array showing (Density).

Did you know?

For visualizations that feature 3D glyphs, it is typically much faster to use the 3D Glyph representation rather than the Glyph filter. This is because the glyph representation draws the same geometry at many different locations (a graphics technique called geometry instancing) while the Glyph filter makes many copies of the same source geometry and renders the resulting mesh in its entirety. Generating all the glyphs and rendering them takes potentially a lot of memory and is typically slower to render, so you should use the 3D Glyph representation when possible.

Next, we will cover each of the property groups available under Display properties. Several of these are marked as advanced. Accordingly, you may need to either toggle the panel to show advanced properties using the button or search for it by name using the search box.

Display properties under Coloring allow you to set how the dataset is colored. To select a single solid color to use to fill the surface or color the wireframe or points, select Solid Color in the combo-box, and then click Edit . That will pop up the standard color chooser dialog from which you can pick a color to use.

If instead you want to pseudocolor using an attribute array available on the dataset, select that array name from the combo-box. For multi-component arrays, you can pick a particular component or Magnitude to use for scalar coloring. ParaView will automatically set up a color transfer function it will use to map the data array to colors. The default range for the transfer function is set up based on the Transfer Function Reset Mode general setting in the Settings dialog when the transfer function is first created. If another dataset is later colored by a data array with the same name, the range of the transfer function will be updated according to the Automatic Rescale Range Mode property in the Color Map Editor . To reset the transfer function range to the range of the data array in the selected dataset, you can use the Rescale button. Remember that, despite the fact that you can set the scalar array with which to color when rendering as Outline , the outline itself continues to use the specified solid color.

Scalar Coloring properties are only relevant when you have selected a data array with which to pseudocolor. The Map Scalars checkbox affects whether a color transfer function should be used (Fig. 4.7). If unchecked, and the data array can directly be interpreted as colors, then those colors are used directly. If not, the color transfer function will be used. A data array can be interpreted as colors if, and only if, it is an unsigned char, float, or double array with two, three, or four components. If the data array is unsigned char, the color values are defined between 0 and 255 while if the data array is float or double, the color values are expected to be between 0 and 1. Interpolate Scalars Before Mapping controls how color interpolation happens across rendered polygons. If on, scalars will be interpolated within polygons, and color mapping will occur on a per-pixel basis. If off, color mapping occurs at polygon points, and colors are interpolated, which is generally less accurate. Refer to the Kitware blog [PatMarion] for a detailed explanation of this option. Use Nan Color For Missing Arrays is a property that, if enabled, will use the special color designated for NaN values in a dataset to also be used as the color for parts of a composite dataset that are missing the scalars array used for color mapping.

Fig. 4.7 The Map Scalars property can be used to avoid using a transfer function and directly interpreting the array values as colors, if possible.

The Polar Axes checkbox toggles polar axes shown around the data. Many parameters can be accessed via an Edit button alongside it. The parameters include angles, tick range, labels, logarithmic mode, ellipse ratio and more.

Fig. 4.8 A Polar Axes usage example.

Styling properties include Opacity (useful when rendering translucent geometries), Point Size (used to control size of points rendered with using Points representation), and Line Width (used to control the thickness of lines when rendering as Wireframe or that of the edges when rendering as Surface With Edges .

Lighting properties affect the shading for rendered surfaces. Interpolation allows you to pick between Flat and Gouraud shading. Specular , together with Specular Color and Specular Power , affects the shininess of the surface. Set this to a non-zero value to render shiny, metallic surfaces.

Common Errors

Specular highlights can lead to misinterpretation of scalar values when using scalar coloring, since the color shown on the shiny part of the surface will not correspond to any color on the color transfer function. Hence, it is generally advisable to use specular highlights on surfaces colored with a single solid color and not on those using scalar coloring (or pseudocoloring).

Edge Styling allows you to set the Edge Color with which to color the edges when using Surface With Edges representation.

Backface Styling provides advanced controls to fine-tune the rendering by controlling front and back faces. A front face is any mesh face facing the camera, while a back face is the one facing away from the camera. By choosing to Cull Frontface or Cull Backface , or by selecting a specific representation type to use for the backface, you can customize your visualizations.

Transforming properties can be used to transform the rendered data in the scene without affecting the raw data itself. Thus, if you apply filters on the data source, it will indeed be working with the untransformed data. To transform the data itself, you should use the Transform filter.

The Coordinate Shift Scale Method is used to choose how to normalize point coordinates to improve rendering quality. Mesh points are sent to the GPU as single-precision float data which can result in resolution issues due to limited precision. VTK includes a variety of methods to normalize the point coordinates to a better range for single-precision floats prior to sending them to the GPU. Auto Shift Scale is a good setting that should work for most datasets - it recomputes a shift and scale factor according to a heuristic involving dataset size and position relative to the origin. Always Auto Shift Scale recomputes the shift and scale every time. Auto Shift Only only shifts the data - this is useful when data is far away from the origin. Near Focal Plane Shift Scale and Focal Point Shift Scale works based on the current camera near clipping point and viewpoint, respectively. This makes it the most robust setting, especially for very large datasets, but it will renormalize the points occasionally as the camera’s settings change. Renormalizing points requires reuploading the data to the GPU, so there may be a performance cost with these last methods.

Several properties are available under the Miscellaneous group. Uncheck the Pickable option if you want the dataset to be ignored when making selections. If the dataset has a texture coordinates array, you can apply a texture to the dataset surface using the Texture combo-box. Choose Load to load a texture or apply a previously loaded texture listed in the combo-box. If your dataset doesn’t have texture coordinates, you can create them by applying one of Texture Map to Cylinder , Texture Map to Sphere , or Texture Map To Plane filters, or using the filters Calculator or Programmable Filter .

The Triangulate option is useful for rendering objects with non-convex polygons. It comes with some additional processing cost for converting polygons to triangles, so it should be used only when necessary.

Fig. 4.9 A dataset made of quadratic tetra hedra displayed with 1, 2, and 3 levels of nonlinear subdivision.

The property Use Shader Replacements enables you to customize the shader code VTK uses for rendering by specifying shader replacements with a JSON string. The JSON string can be a single node or an array of nodes with the following properties:

• “type”: specifies the type of shader the replacement is about. It can be either “vertex”, “fragment” or “geometry”.

• “original”: specifies the original string to be replaced in the shader code. This string is generally a pattern defined by the mapper vtkOpenGLPolyDataMapper at specific locations of the shader GLSL source code.

• “replacement”: specifies the replacement string in GLSL source code. Note that the Json parser supports multiple lines entries.

Here’s an example of a simple shader replacement (draw all the fragments in full red color without any shading consideration):

{
  "type": "fragment",
  "original": "//VTK::Light::Impl",
  "replacement": "gl_FragData[0]=vec4(1.0,0.0,0.0,1.0);"
}

The Nonlinear Subdivision Level property is used when rendering datasets with higher- order elements. Use this to set the subdivision level for triangulating higher order elements. The higher the value, the smoother the edges. This comes at the cost of more triangles and, hence, potentially, increased rendering time.

The Block Colors Distinct Values property sets the number of unique colors to use when coloring multiblock datasets by block ID. Finally, Use Data Partitions controls whether data is redistributed when it is rendered translucently. When off (default value), data is repartitioned by the compositing algorithm prior to rendering. This is typically an expensive operation that slows down rendering. When this option is on, the existing data partitions are used, and the cost of data restribution is avoided. However, if the partitions are not sortable in back-to-front order, rendering artifacts may occur.

Volume Rendering options are available if the data can be volume rendered. You can pick a specific type of Volume Rendering Mode , although the default ( Smart ) should work in most cases, since it attempts to pick a volume rendering mode suitable for your data and graphics setup. To enable gradient-based shading, check Shade , if available.

Slicing properties are available when the Slice representation type is present. These allow you to pick the orthogonal slice plane orientation and slice offset using Slice Direction and the Slice slider.

4.4.3. Render View in pvpython

4.4.3.1. Creating a Render View

You use CreateRenderView or CreateView functions to create a new instance of a render view.

>>> from paraview.simple import *
>>> view = CreateRenderView()
# Alternatively, use CreateView.
>>> view = CreateView("RenderView")

noindent You use Show and Hide to show or hide data produced by a pipeline module in the view.

>>> source = Sphere()
>>> view = CreateRenderView()

# Show active source in active view.
>>> Show()

# Or specify source and view explicitly.
>>> Show(source, view)

# Hide source in active view.
>>> Hide(source)

4.4.3.2. Interactions

Since pvpython is designed for scripting and batch processing, it has limited support for direct interaction with the view. To interact with a scene, invoke the Interact function in Python.

Interact()

More often, you will programmatically change the camera as follows:

# Get camera from the active view, if possible.
>>> camera = GetActiveCamera()

# or, get the camera from a specific render view.
>>> camera = view.GetActiveCamera()

# Now, you can use methods on camera to move it around the scene.

# Divide the camera's distance from the focal point by the given dolly value.
# Use a value greater than one to dolly-in toward the focal point, and use a
# value less than one to dolly-out away from the focal point.
>>> camera.Dolly(10)

# Set the roll angle of the camera about the direction of projection.
>>> camera.Roll(30)

# Rotate the camera about the view up vector centered at the focal point. Note
# that the view up vector is whatever was set via SetViewUp, and is not
# necessarily perpendicular to the direction of projection. The result is a
# horizontal rotation of the camera.
>>> camera.Azimuth(30)

# Rotate the focal point about the view up vector, using the camera's position
# as the center of rotation. Note that the view up vector is whatever was set
# via SetViewUp, and is not necessarily perpendicular to the direction of
# projection. The result is a horizontal rotation of the scene.
>>> camera.Yaw(10)

# Rotate the camera about the cross product of the negative of the direction
# of projection and the view up vector, using the focal point as the center
# of rotation. The result is a vertical rotation of the scene.
>>> camera.Elevation(10)

# Rotate the focal point about the cross product of the view up vector and the
# direction of projection, using the camera's position as the center of
# rotation. The result is a vertical rotation of the camera.
>>> camera.Pitch(10)

Alternatively, you can explicitly set the camera position, focal point, view up, etc,. to explicitly place the camera in the scene.

>>> camera.SetFocalPoint(0, 0, 0)
>>> camera.SetPosition(0, 0, -10)
>>> camera.SetViewUp(0, 1, 0)
>>> camera.SetViewAngle(30)
>>> camera.SetParallelProjection(False)

# If ParallelProjection is set to True, then you'll need
# to specify parallel scalar as well i.e. the height of the viewport in
# world-coordinate distances. The default is 1. Note that the `scale'
# parameter works as an `inverse scale' where larger numbers produce smaller
# images. This method has no effect in perspective projection mode.
>>> camera.SetParallelScale(1)

4.4.3.3. View properties

In pvpython, view properties are directly accessible on the view object returned by CreateRenderView or GetActiveView .

Once you get access to the view properties objects, you can then set properties on it similar to properties on pipeline modules such as sources, filters, and readers.

>>> view = GetActiveView()

# Set center axis visibility
>>> view.CenterAxesVisibility = 0

# Or you can use this variant to set the property on the active view.
>>> SetViewProperties(CenterAxesVisibility=0)

# Another way of doing the same
>>> SetViewProperties(view, CenterAxesVisibility=0)

# Similarly, you can change orientation axes related properties
>>> view.OrientationAxesVisibility = 0
>>> view.OrientationAxesLabelColor = (1, 1, 1)

4.4.3.4. Display properties

Similar to view properties, display properties are accessible from the display properties object or using the SetDisplayProperties function.

>>> displayProperties = GetDisplayProperties(source, view)
# Both source and view are optional. If not specified, the active source
# and active view will be used.

# Now one can change properties on this object
>>> displayProperties.Representation = "Outline"

# Or use the SetDisplayProperties API.
>>> SetDisplayProperties(source, view, Representation=Outline)

# Here too, source and view are optional and when not specified,
# active source and active view will be used.

You can always use the help function to get information about available properties on a display properties object.

4.5. Line Chart View

Fig. 4.10 paraview using Line Chart View to plot data values probed along a line through the dataset using Plot Over Line filter.

Line Chart View can be used to plot data as a line plot representing changes in dependent variables against an independent variable. Using display properties, you can also show scatter plots in this view. This view and other charting views in ParaView follow a similar design, where you pick attribute arrays to plot using display properties, and they are plotted in the view. How those values get plotted depends on the type of the view: Line Chart View draws a line connecting sample points, Bar Chart View renders bars at each sample point, etc.

One of the most common ways of showing a line plot is to apply the Plot Over Line filter to any dataset. This will probe the dataset along the probe line specified. You then plot the sampled values in the Line Chart View . Alternatively, if you have a tabular dataset (i.e. vtkTable ), then you can directly show the data in this view.

Did you know?

You can plot any arbitrary dataset, even those not producing vtkTable outputs, by using the Plot Data filter. Remember, however, that for extremely large datasets, while Render View may use parallel rendering strategies to improve performance and reduce memory requirements, chart views rarely, if ever, support such parallel strategies.

4.5.1. Understanding plotting

Line Chart View plots data arrays. For any dataset being shown in the view, you first select which data array is to be treated as the independent variable and plotted along the x-axis. Then, you select which arrays to plot along the Y-axis. You can select multiple of these and setup properties for each of the series so they are rendered with different colors and line styles. Since data arrays in VTK datasets are associated with cells or points, and the two are not directly comparable to one another, you can only pick arrays associated with one type of attribute at any time.

4.5.2. Line Chart View in paraview

4.5.2.1. Creating a Line Chart View

Similar to creating Render View , you can split the viewport or convert an existing view to Line Chart View . Line Chart View will also be automatically created if you apply a filter that needs this view, e.g., the Plot Over Line filter.

Did you know?

If you generate lengthy data for the Line Chart View , the default variables that are selected may be slow to adjust. You can change paraview’s default behavior to initially load no variables at all by selecting the Load No Chart Variables checkbox under Settings/General/Properties Panel Options .

4.5.2.2. Interactions

Interactions with the chart view result in changing the plotted axes ranges. You can left-click and drag to pan, i.e., change the origin. To change the range on either of the axes, you can right-click and drag vertically and/or horizontally to change the scale on the vertical axes and/or horizontal axes, respectively.

You can also explicitly specify the axes range using view properties.

4.5.2.3. View properties

The view properties for Line Chart View are grouped as properties that affect the view and those that affect each of the potential four axes.

To set a title, use Chart Title . Title text properties such as font, size, style, and alignment with respect to the chart can be set under Chart Title Properties . To toggle the visibility of the legend, use Show Legend . While you cannot interactively place the legend in this view, you can use Legend Location to place it at one of the corners.

There are four axes possible in this view: left, bottom, top, and right. The top and right axes are shown only when some series is set to use those. (We will cover this in the Display properties subsection.) For each of the axes, you can set a title (e.g., Left Axis Title , Bottom Axis Title , etc.) and adjust the title font properties. You can turn on a grid with a customizable color by checking the Show Left Axis Grid , for example.

Next, you can customize the axes ranges. You can always simply interact with the mouse to set the axes ranges. To precisely set the range, check the Axis Use Custom Range for the appropriate axis, e.g., Bottom Axis Use Custom Range for fixing the bottom axis range, and then specify the data values to use for the min and the max.

The labels on the axes are, by default, automatically determined to avoid visual clutter. By default, the axis labels are arranged on a linear scale, but by enabling the Axis Log Scale option you can use log scaling instead. In addition, you can override the default labelling strategy for any of the axes separately and, instead, specify the locations to label explicitly. This can be done by checking Axis Use Custom Labels for a particular axis, e.g., Bottom Axis Use Custom Labels . When checked, a list widget will be shown where you can manually add values at which labels will be placed.

For generating log plots, simply check the corresponding Axis Use Log Scale , e.g., Left Axis Use Log Scale to use log scale for Y-axis (Fig. 4.11). Note that log scale should only be used for an axis with a non-zero positive range, since the log of a number less than or equal to 0 is undefined.

Fig. 4.11 Differences between line charts when using log scale for the Y-axis.

4.5.2.4. Display properties

Display properties allow you to setup which series or data arrays are plotted in this view. You start by picking the Attribute Type . Select the attribute type that has the arrays of interest. For example, if you are plotting arrays associated with points, then you should pick Point Data .) Arrays with different associations cannot be plotted together. You may need to apply filters such as Cell Data to Point Data or Point Data to Cell Data to convert arrays between different associations to do that.

Properties under X Axis Parameters allow you to select the independent variable plotted on the X axis by choosing the X Array Name . If none of the arrays are appropriate, you can choose to use the element index in the array as the X axis by checking Use Index for XAxis .

Series Parameters control series or data arrays plotted on the Y-axis. All available data arrays are lists in the table widget that allows you to check/uncheck a series to plot in the first column. The second column in the table shows the associated color used to plot that series. You can double-click the color swatch to change the color to use. By default, ParaView will try to pick a palette of discrete colors. The third column lets you set the opacity of the series plot elements. The fourth column (Variable) shows the name of the variable to plot. The fifth column (Legend Name) shows the label to use for that series in the legend. By default, it is set to be the same as the array name. You can double-click to change the name to your choice, e.g., to add units.

Other series parameters include Line Thickness, Line Style, Marker Style, and Marker Size. To change any of these, highlight a row in The Series Parameters widget, and then change the associated parameter to affect the highlighted series. You can change properties for multiple series and can select multiple of them by using the CTRL (or ⌘) and ⇧ keys.

Using Chart Axes , you can change which axes on which a series is shown. The default is Bottom-Left , but you can change it to be Bottom-Right , Top-Left , or Top-Right to accommodate series with widely different ranges in the same plot.

4.5.3. Line Chart View in pvpython

The principles involved in accessing Line Chart View from pvpython are similar to those with Render View . You work with the view properties and display properties objects to change views and display properties, respectively. The thing that changes is the set of available properties.

The following script demonstrates the typical usage:

>>> from paraview.simple import *

# Create a data source to probe into.
>>> Wavelet()
<paraview.servermanager.Wavelet object at 0x1156fd810>

# We update the source so that when we create PlotOverLine filter
# it has input data available to determine good defaults. Otherwise,
# we will have to manually set up the defaults.
>>> UpdatePipeline()

# Now, create the PlotOverLine filter. It will be initialized using
# defaults based on the input data.
>>> PlotOverLine()
<paraview.servermanager.PlotOverLine object at 0x1156fd490>

# Show the result.
>>> Show()
<paraview.servermanager.XYChartRepresentation object at 0x1160a6a10>

# This will automatically create a new Line Chart View if the
# the active view is no a Line Chart View since PlotOverLine
# filter indicates it as the preferred view. You can also explicitly
# create it by using CreateView() function.

# Display the result.
>>> Render()

# Access display properties object.
>>> dp = GetDisplayProperties()
>>> print(dp.SeriesVisibility)
['arc_length', '0', 'RTData', '1']

# This is  list with key-value pairs where the first item is the name
# of the series, then its visibility and so on.
# To toggle visibility, change this list e.g.
>>> dp.SeriesVisibility = ['arc_length', '1', 'RTData', '1']

# Same is true for other series parameters including series color,
# line thickness etc.

# For series color, the value consists of 3 values: red, green, and blue
# color components.
>>> print(dp.SeriesColor)
['arc_length', '0', '0', '0', 'RTData', '0.89', '0.1', '0.11']

# For series labels, value is the label to use.
>>> print(dp.SeriesLabel)
['arc_length', 'arc_length', 'RTData', 'RTData']

# e.g. to change RTData's legend label, we can do something as follows:
>>> dp.SeriesLabel[3] = 'RTData -- new label'

# Access view properties object.
>>> view = GetActiveView()
# or
>>> view = GetViewProperties()

# To change titles
>>> view.ChartTitle = "My Title"
>>> view.BottomAxisTitle = "X Axis"
>>> view.LeftAxisTitle = "Y Axis"

4.6. Bar Chart View

Fig. 4.12 paraview using Bar Chart View to plot the histogram for a dataset using the Histogram filter.

Bar Chart View is very similar to Line Chart View when it comes to creating the view, view properties, and display properties. One difference is that, instead of rendering lines for each series, this view renders bars. In addition, under the display properties, the Series Parameters like Line Style and Line Thickness are not available, since they are not applicable to bars.

4.7. Box Chart View

Fig. 4.13 paraview using Box Chart View to plot the box chart for a dataset using the Compute Quartiles filter.

Box plot is a standard method for graphically depicting groups of statistical data based on their quartiles. A box plot is represented by a box with the following properties: the bottom of the rectangle corresponds to the first quartile, a horizontal line inside the rectangle indicates the median and the top of the rectangle corresponds to the third quartile. The maximum and minimum values are depicted using vertical lines that extend from the top and the bottom of the rectangle.

In ParaView, the Box Chart View can be used to display such box plots through the Compute Quartiles filter which computes the statistical data needed by the view to draw the box plots.

4.8. Plot Matrix View

Fig. 4.14 paraview using Plot Matrix View to generate a scatter plot matrix to understand correlations between pairs of variables.

Plot Matrix View is a chart view that renders a scatter plot matrix. It allows you to spot patterns in the small scatter plots, change focus to those plots of interest, and perform basic selection. The principle is that, for all selected arrays or series to be plotted, the view generates a scatter plot for each pair. You can activate a particular scatter plot, in which case the active plot is re-drawn at a bigger scale to make it easier to inspect. Additionally, the view shows a histogram for each plotted variable or series.

The view properties allow you to set colors to use for active plot, histograms, etc., while the display properties allow you to pick which series are plotted.

4.8.1. Interactions

You can click on any of the plots (except the histograms) in the matrix to make it active. Once actived, the active plot will show that plot. You can then interact with the active plot exactly like Line Chart View or Bar Chart View for panning and zoom.

4.8.2. View properties

View properties on this view allow you to pick styling parameters for the rendering ranging from title ( Chart Title ) to axis colors ( Active Plot Axis Color , Active Plot Grid Color ). You can also control the visibility of the histrogram plots, the active plot, the axes labels, the grids, and so on.

4.8.3. Display properties

Similar to Line Chart View , you select the Attribute Type and then the arrays to plot. Since, in a scatter plot matrix, the order in which the selected series are rendered can make it easier to understand correlations, you can change the order by clicking and dragging the rows in the Series Parameters table.

4.9. Parallel Coordinates View

Fig. 4.15 paraview using Parallel Coordinates View to plot correlations between columns in a table.

Like Plot Matrix View , Parallel Coordinates View is also used to visualize correlations between data arrays.

One of the main features of this view is the ability to select specific data in order to analyse the factors influencing the data. e.g., with a table of three variables, one being the “output” variable the other two being the potential factor influencing the first, selecting only the output will enable you to see if none, one, or both of the factors are actually influencing the output.

Fig. 4.16 High “S” data point are influenced more by low “Ks” than high “Q”

4.10. Spreadsheet View

Fig. 4.17 paraview using SpreadSheet View to plot raw data values for the can.ex2 dataset.

SpreadSheet View is used to inspect raw data values in a tabular form. Unlike most other views, this view is primarily intended to be used in the paraview user interface and, hence, is not available in pvpython.

To use this view, simply create this view and show the dataset produced by any pipeline module in it by using the Pipeline Browser . SpreadSheet View can only show one dataset at a time. Therefore, showing a new dataset will automatically hide the previously shown dataset.

The view’s toolbar provides quick access to several of this view’s capabilities. Use the Showing widget on the view toolbar to view as well as to change the dataset being shown. The Attribute field allows you to pick which types of elements to show, e.g., Cell Data , Point Data , Field Data , etc. Precision can be utilized to change the precision used when displaying floating point numbers. The button enables you to select columns to show. Click on the button to get a popup menu in which you check/uncheck the columns to show/hide. If showing Cell Data , the button, when checked, enables you to see the point ids that form each of the cells.

Section 6.1 discusses how selections can be made in views to select elements of interest. Use the button to make the view show only selected elements. Now, as you make selections in other views, this SpreadSheet View will update to only show the values of the selected elements (as long as the dataset selected in are indeed being shown in the view).

Did you know?

Some filters and readers produce vtkTable, and they are automatically displayed in a spreadsheet view. Thus, one can very easily read the contents of a .csv file with ParaView.

4.11. Slice View

Fig. 4.18 Slice View can be used to show orthogonal slices from datasets.

Slice View is special type of Render View that can be used to view orthogonal slices from any dataset. Any dataset shown in the view will be sliced in axis-aligned slices based on the locations specified on the view. The slice locations along the three orthogonal axis planes can be specified by using the frame decoration around the view window.

4.11.1. Interactions

Since this view is a type of Render View , the camera interactions are same as that of Render View . Additionally, you can interact with the frame decoration to manipulate slice locations along the three axis planes.

• Double-click the left mouse button in the region between the axis border and the view to add a new slice.

• You can click-and-drag a marker to move the slice location.

• To remove a slice, double-click with the left mouse button on the marker corresponding to that slice.

• To toggle visibility of the slice, you can right-click on the marker.

4.11.2. Slice View in pvpython

# To create a slice view in use:
>>> view = CreateView("MultiSlice")

# Use properties on view to set/get the slice offsets.
>>> view.XSliceValues = [-10, 0, 10]
>>> print(view.XSliceValues)
[-10, 0, 10]

# Similar to XSliceValues, you have YSliceValues and ZSliceValues.
>>> view.YSliceValues = [0]
>>> view.ZSliceValues = []

4.12. Python View

Some Python libraries, such as matplotlib, are widely used for making publication-quality plots of data. The Python View provides a way to display plots made in a Python script right within paraview.

The Python View has a single property, a Python script that generates the image to be displayed in the viewport. All the Python bindings for ParaView and VTK that are available in the Python scripting module are available from within this script, making it possible to plot any array from just about any dataset that can be loaded into ParaView. The Python script for the view is evaluated in a unique Python environment so that global variables defined in the script do not clobber global variables in other Python scripts (either in other instances of the Python View or in the Python interpreter). This environment is reset each time the script is evaluated, so data cannot be saved between evaluations.

The Python View requires that the Python script where the plotting occurs define two functions. In the first function, you request which arrays you would like to transfer to the client for rendering. At present, all rendering in this view takes place on the client, even in client-server mode. These arrays can be point data, cell data, field data, and table row data. This function runs only on data-server processes. It provides access to the underlying data object on the server so that you can query any aspect of the data using the Python-wrapped parts of VTK and ParaView.

The second function is where you put Python plotting or rendering commands. This function runs only on the ParaView client. It has access to the complete data object gathered from the data server nodes, but only has access to the arrays requested in the first function. This function will typically set up objects from a plotting library, convert data from VTK to a form that can be passed to the plotting library, plot the data, and convert the plot to an image (a vtkImageData object) that can be displayed in the viewport.

4.12.1. Selecting data arrays to plot

All the rendering in the Python View occurs in the client, so the client must get the data from the server. Because the dataset residing on the ParaView server may be large, transferring all the data to the client may not be possible or practical. For that reason, we have provided a mechanism to select which data arrays in a data object on the server to transfer to the client. The overall structure of the data object, however, (including cell connectivity, point positions, and hierarchical block structure) is always transferred to the client. By default, no data arrays are selected for transfer from the server.

The Python script for the view must define a function called setup_data(view) . The view argument is the VTK object for the Python View . The current datasets loaded into ParaView may be accessed through the view object.

Fig. 4.19 paraview using Python View with matplotlib to display a scatterplot of momentum magnitude versus density (upper right) and a histogram of density (lower right) in the bluntfin.vts dataset.

Here’s an example of this function that was used to generate the image in Fig. 4.19:

def setup_data(view):
  # Iterate over visible data objects
  for i in range(view.GetNumberOfVisibleDataObjects()):
    # You need to use GetVisibleDataObjectForSetup(i)
    # in setup_data to access the data object.
    dataObject = view.GetVisibleDataObjectForSetup(i)

    # The data object has the same data type and structure
    # as the data object that sits on the server. You can
    # query the size of the data, for instance, or do anything
    # else you can do through the Python wrapping.
    print('Memory size: {0} kilobytes'.format(dataObject.GetActualMemorySize()))

    # Clean up from previous calls here. We want to unset
    # any of the arrays requested in previous calls to this function.
    view.DisableAllAttributeArrays()

    # By default, no arrays will be passed to the client.
    # You need to explicitly request the arrays you want.
    # Here, we'll request the Density point data array
    view.SetAttributeArrayStatus(i, vtkDataObject.POINT, "Density", 1)
    view.SetAttributeArrayStatus(i, vtkDataObject.POINT, "Momentum", 1)

    # Other attribute arrays can be set similarly
    view.SetAttributeArrayStatus(i, vtkDataObject.FIELD, "fieldData", 1)

The vtkPythonView class passed in as the view argument to setup_data(view) defines several methods useful for specifying which data arrays to copy:

• GetNumberOfVisibleDataObjects() - This returns the number of visible data objects in the view. If an object is not visible, it should not show up in the rendering, so all the methods provided by the view deal only with visible objects.

• GetVisibleDataObjectForSetup(visibleObjectIndex) - This returns the visibleObjectIndex'th visible data object in the view. (The data object will have an open eye next to it in the Pipeline Browser .)

• GetNumberOfAttributeArrays(visibleObjectIndex, attributeType) - This returns the number of attribute arrays for the visibleObjectIndex'th visible object and the given attributeType (e.g., vtkDataObject.POINT , vtkDataObject.CELL , etc.).

• GetAttributeArrayName(visibleObjectIndex, attributeType, arrayIndex) - This returns the name of the array of the given attribute type at the given array index for the visibleObjectIndex'th object.

• SetAttributeArrayStatus(visibleObjectIndex, vtkDataObject.POINT, "Density", 1) - This sets the array status of an attribute array. The first argument is the visible object index, the second object is the attribute association of the array, the third argument is the name of the array, and the last argument specifies if the array is to be copied (1) or not (0).

• GetAttributeArrayStatus(visibleObjectIndex, vtkDataObject.POINT, "Density") - This retrieves the array status for the object with the given visible index with a given attribute association (second argument) and a name (last argument).

• EnableAllAttributeArrays() - This sets all arrays to be copied.

• DisableAllAttributeArrays() - This sets all arrays to not be copied.

The methods GetNumberOfVisibleDataObjects() , GetVisibleDataObjectForSetup(...) , GetNumberOfAttributeArrays(...) , and GetAttributeArrayName(...) are all convenient methods for obtaining information about visible data objects in the view that could otherwise be accessed with existing view and representation methods. The last four methods are valid only from within the setup_data(view) function.

4.12.2. Plotting data in Python

After the setup_data(view) function has been called, ParaView will transfer the data object and selected arrays to the client. When that is done, it will call the render(view, width, height) function you have defined in your script.

The view argument to the render(view, width, height) function is the vtkPythonView object on the client. The width and height arguments are the width and height of the viewport, respectively. The render(view, width, height) function uses the data available through the view, along with the width and height, to generate a vtkImageData object that will be displayed in the viewport. This vtkImageData object must be returned from the render(view, width, height) function. If no vtkImageData is returned, the viewport will be black. If the size of the image does not match the size of the viewport, the image will be stretched to fit the viewport.

Putting it all together, here is a simple example that generates a solid red image to display in the viewport.

def render(view, width, height):
  from paraview.vtk import vtkImageData
  image = vtkImageData()
  image.SetDimensions(width, height, 1)
  from paraview.numeric import VTK_UNSIGNED_CHAR
  image.AllocateScalars(VTK_UNSIGNED_CHAR, 4)
  pixel_array = image.GetPointData().GetArray(0)
  pixel_array.FillComponent(0, 255.0)
  pixel_array.FillComponent(1, 0.0)
  pixel_array.FillComponent(2, 0.0)
  pixel_array.FillComponent(3, 0.0)

  return image

This example does not produce an interesting visualization, but serves as a minimal example of how the render(view, width, height) function should be implemented. Typically, we expect that the Python plotting library you use has some utilities to expose the generated plot image pixel data. You need to copy that pixel data to the vtkImageData object returned by the render(view, width, height) function. Exactly how you do this is up to you, but ParaView comes with some utilities to make this task easier for matplotlib.

4.12.2.1. Set up a matplotlib Figure

The Python View comes with a Python module, called python_view , that has some utility functions you can use. To import it, use:

from paraview import python_view

This module has a function, called matplotlib_figure(view, width, height) , that returns a matplotlib.figure.Figure given width and height arguments. This figure can be used with matplotlib plotting commands to plot data as in the following:

def render(view, width, height):
  figure = python_view.matplotlib_figure(width, height)

  ax = figure.add_subplot(1,1,1)
  ax.minorticks_on()
  ax.set_title('Plot title')
  ax.set_xlabel('X label')
  ax.set_ylabel('Y label')

  # Process only the first visible object in the pipeline browser
  dataObject = view.GetVisibleDataObjectForRendering(0)

  x = dataObject.GetPointData().GetArray('X')

  # Convert VTK data array to numpy array for plotting
  from paraview.numpy_support import vtk_to_numpy
  np_x = vtk_to_numpy(x)

  ax.hist(np_x, bins=10)

  return python_view.figure_to_image(figure)

This definition of the render(view, width, height) function creates a histogram of a point data array named X from the first visible object in the Pipeline Browser . Note the conversion function, python_view.figure_to_image(figure) , in the last line. This converts the matplotlib Figure object created with python_view.matplotlib_figure(width, height) into a vtkImageData object suitable for display in the viewport.

4.13. Comparative Views

Comparative Views , including Render View (Comparative) , Line Chart View (Comparative) , and Bar Chart View (Comparative) , are used for generating comparative visualization from parameter studies. These views are covered in Section 4 of the Reference Manual.

4.14. Virtual reality

ParaView supports immersive data visualization in virtual reality (VR) with head-mounted displays (HMDs). VR can be used to view the scene in the active Render View and interact with it. This environment provides enhanced perception of depth and more intuitive manipulation of 3D objects by creating the sensation of the user being physically present inside of the scene.

4.14.1. XRInterface plugin

VR support in ParaView is provided through the XRInterface plugin (previously called OpenVR plugin). To load this plugin, open the Plugin Manager via Tools > Manage Plugins…. Click on XRInterface, then on the button Load Selected. This will open the XRInterface panel.

4.14.1.1. How to install/build

By default, the XRInterface plugin is already enabled in ParaView Windows binaries. Simply launch ParaView with a VR runtime such as SteamVR on your machine to easily experience VR with ParaView.

Otherwise, the plugin needs to be enabled before building ParaView, see README.md in the plugin sources for more information.

4.14.1.2. XRInterface panel

By default, the XRInterface panel appears on the right upon loading the plugin. To open it manually, search for the corresponding checkbox in the main menu via View > XRInterface.

Fig. 4.20 XRInterface panel.

Once the XRInterface plugin is loaded and an HMD connected, click on the button Send to XR to start rendering in VR. Other options are available in the plugin panel:

• Use MultiSamples — This checkbox indicates whether multisampled framebuffers are used. Using multisamples can reduce flickering but currently does not work with volume rendering.

• Base Station Visibility — This checkbox indicates whether the base stations models are displayed in VR (for devices using base stations).

• Desired XR Runtime — This drop down list indicates whether to use OpenVR or OpenXR when both are available.

• Send to XR - This button launches the VR mode using the selected runtime.

• Attach to Current View — This button allows VR-only elements to appear in the current Render View, such as cropping planes or avatars. This option is mainly useful in Collaboration mode for desktop users when an HMD is not available.

• Show XR View - This button opens a window showing a stabilized view of what the VR headset is showing, which is useful to produce screenshots or videos.

• Export Locations as a View — This button exports saved locations as .vtp files in a folder called pv-view. This option is meant to be used for Mineview.

• Export Locations as Skyboxes — This button exports saved locations as skyboxes in a folder called pv-skybox. This generates six .jpg images (right, left, up, down, back and front) producing a skybox that can then be used outside of VR as well.

4.14.2. Interactions

On top of head tracking to reproduce the user physical movements for navigation, the controllers can be used to interact with the data through actions such as scaling, picking, etc.

The controls mapping for different controllers is detailed in Section 4.14.5.

Action	Controls
Movement	(Left/Right) Joystick
Open XR Menu	Right Menu Button
Confirm (XR Menu)	Right Trigger
Trigger Action	Right Trigger
Move to Next Saved Location	Left Trigger
Grip Action	Left Grip + Right Grip

4.14.2.1. Trigger actions

Trigger actions are assigned to the right trigger by default and include grabbing, picking, probing, interactive clipping, teleportation, and adding points to sources (such as a polyline source). The current action can be chosen via the XR menu (see Section 4.14.3).

• Adding points to a source — Press the right trigger to place a point at the tip of the right controller. Only valid when the active source allows placing points, such as a polyline source.

  

  Fig. 4.21 Adding a point to a polyline source in VR.

• Grabbing — Press the right trigger when the right controller is in the bounding box of an object to grab it and move it.

  

  Fig. 4.22 Grab action in VR.

• Picking — Press the right trigger while pointing at an object to pick it and move it from a distance.

  

  Fig. 4.23 Picking action in VR.

• Interactive Cropping — Press the right trigger to crop the scene in real time with a plane placed at the tip of the right controller.

  

  Fig. 4.24 Real time cropping in VR.

• Probing — Press the right trigger while pointing at a dataset cell to select it and display information on its position and data values.

  

  Fig. 4.25 Cell probing action in VR.

• Teleportation — Press the right trigger while pointing at a dataset with the ray to instantly move at the pointed location.

4.14.2.2. Grip actions

Grip actions are used to transform the scene through translation, scaling and rotation.

• Translation — Press both grip buttons, then move the controllers together to translate the scene.

  

  Fig. 4.26 Translation action in VR. Green and red arrows correspond to movements of the controllers and dataset, respectively.

• Scaling/zoom — Press both grip buttons, then move the controllers farther or closer to each other to zoom in or out of the scene.

  

  Fig. 4.27 Scaling action in VR. Green and red arrows correspond to movements of the controllers and dataset, respectively.

• Rotation — Press both grip buttons, then move one controller around the other to rotate the scene around a vertical axis.

  

  Fig. 4.28 Rotation action in VR. Green and red arrows correspond to movements of the controllers and dataset, respectively.

4.14.3. XR menu

In the XR View, pressing the menu button opens a dedicated menu floating in the 3D scene in front of the user.

Fig. 4.29 XR integrated menu.

This menu is composed of the following elements:

1. Pipeline Browser — This is the same Pipeline Browser present in ParaView. The visibility for each item in the pipeline can be modified by pointing the navigation ray on the eye icon and pressing the right trigger.

2. Panels — VR options are distributed into 4 panels, that can be displayed by clicking on the corresponding tab:

   • Interaction — This panel contains options related to the interactions with the scene using the controllers (see Section 4.14.3.1).

   • Movement — This panel contains options related to the camera movement and poses (see Section 4.14.3.2).

   • Environment — This panel contains global options related to the scene (see Section 4.14.3.3).

   • Widgets — This panel contains options related to VR-specific widgets (see Section 4.14.3.4).

3. Exit XR — This button closes the current XR View.

4. Animation Buttons — These buttons are used to navigate timesteps for temporal datasets.

4.14.3.1. Interaction panel

Fig. 4.30 Interaction panel of the XR integrated menu.

• Right Trigger — This drop down list indicates which action is mapped to the right trigger outside of the menu, such as picking, probing, etc. See Section 4.14.2.1 for the list of available actions.

• Reset Actors Positions — This button resets the position of all objects within the scene.

• Interactive Ray — This checkbox indicates whether the ray changes color when pointing at an object to signal a collision.

4.14.3.2. Movement Panel

Fig. 4.31 Movement panel of the XR integrated menu.

• Movement Style — This drop down list indicates which movement style is mapped to the controller joysticks.

  • Flying mode uses only one joystick and movement follows the orientation of the controller itself.

  • Grounded mode uses the left joystick to move horizontally in all directions, while the right joystick controls elevation.

• Movement Speed — This horizontal slider modifies the movement speed when using the joysticks. A higher multiplier corresponds to a higher speed.

• Reset Camera — This button scales and places the objects at the center of the physical room.

• Camera pose buttons:

  • Clear — This button clears all previously saved camera poses.

  • Save pose — This button saves the current pose in the list of saved camera poses. Up to 6 poses can be saved this way. For each saved pose, a dedicated button is added to the right of this button.

• Bring Collaborators Here — This button moves all collaborators to your current location. Only applicable in Collaboration mode.

4.14.3.3. Environment panel

Fig. 4.32 Environment panel of the XR integrated menu.

• View Up Direction — These buttons set which axis points upwards from the top of the HMD. This is useful when datasets or skyboxes are oriented differently from the default.

• Scene Scale — These buttons change the scaling factor of the scene. A higher value results in all objects appearing larger.

• Show Floor — This button allows hiding or showing the floor as a white plane.

4.14.3.4. Widgets panel

Fig. 4.33 Widgets panel of the XR integrated menu.

• Distance Widget — This button adds a measuring tool to the scene. Press the right trigger once to place the starting point where the right controller is located, then press a second time with the controller at the desired location to place the second point. Four values are displayed next to the tool: distance and X, Y, Z difference between both points. The tips of the line can be grabbed and moved individually after placing them.

Fig. 4.34 Distance measurement widget in VR.

• Navigation Panel — This button allows to display a tooltip above the left controller with information on its spatial position in world coordinates.

Fig. 4.35 Navigation panel widget in VR.

• Cropping buttons — The following buttons provide tools to crop data in real time. Cropping planes can be moved by placing the right controller on them and grabbing them with the right trigger. More than one plane can be added to the scene.

  • Add Crop Plane — This button adds a cropping plane to the scene.

  • Add Thick Crop — This button adds a thick cropping plane to the scene.

  • Hide Crop Planes — This button hides all cropping planes in the scene.

  • Remove All Crop Planes — This button removes all cropping planes from the scene.

  • Crop Thickness — This horizontal slider sets the thickness of created thick cropping planes (this parameter does not affect current ones). By default, the value is set to auto, which adjusts the plane thickness according to the current scene scale.

  • Snap Crop Planes — This button allows to choose whether the cropping planes should snap to the coordinates axes.

4.14.4. Remoting

Notice

Remoting support is currently experimental.

Remoting allows streaming the active render view of ParaView to a remote device.

4.14.4.1. Prerequisites

The remoting feature is only available on Windows for the Hololens 2 and requires an additional package named Microsoft.Holographic.Remoting.OpenXr. With this, ParaView can connect to another application in the remote device if both applications use the same version of this package.

Note that the ParaView release uses the same version as the official player application developed by Microsoft, available in the Microsoft Store, which is version 2.9.2.

If you do have not an application already deployed in the remote device, we recommend downloading the Holographic Remoting Player application in the Microsoft Store.

4.14.4.2. How to launch

First, start the application on the remote device.

After launching this application, it will wait for another application to connect to it via an IP address.

Fig. 4.36 Remote application awaiting connection in the Hololens 2.

You can now start ParaView and do any process on your data that you want. When you are ready to test it in the Hololens 2, enable the XR Interface plugin. You will need to set different options:

• Desired XR Runtime — set it to OpenXR because the Microsoft.Holographic.Remoting.OpenXr depends on it.

• Use OpenXR Remoting — enable or disable the remoting support.

• Remote address — set the IP address to connect ParaView and the application in the Hololens 2.

Fig. 4.37 XRInterface panel with OpenXR Remoting options.

After setting these options, you can click on Send To XR. Once the connection is established, you will be able to see and interact with your dataset.

Fig. 4.38 can.ex2 in the Hololens 2.

For now, only grabbing feature is available as interaction.

4.14.5. Controls mapping

This section details the button mappings used in ParaView for different types of controllers. The controls names are those used throughout the VR section.

4.14.5.1. HP Motion

XR View Controls	Controller Button
Joystick	Joystick
Trigger	Trigger
Grip	Grip Button
Menu Button	Menu Button

4.14.5.2. HTC Vive

XR View Controls	Controller Button
Joystick	Trackpad
Trigger	Trigger
Grip	Grip Button
Menu Button	Menu Button

4.14.5.3. Oculus Touch

XR View Controls	Controller Button
Joystick	Joystick
Trigger	Trigger
Grip	Grip Button
Menu Button	Menu Button

4.14.5.4. Valve Index

XR View Controls	Controller Button
Joystick	Joystick
Trigger	Trigger
Grip	A Button
Menu Button	B Button

4.14.5.5. Microsoft Hand

XR View Controls	Controller Gesture
Trigger	Hand squeeze
Grip	Hand grip