Representations and Views

ParaView's display pipeline supports a huge number of rendering modes and options.

• The output of every filter can be displayed in one or many windows, including arrays of comparative views in which the pipeline is allowed to vary.
• The rendering modalities and visible characteristics vary widely, from text mode to 2D plots, to volume rendering.
• Depending on ParaView's configuration that data can be generated rendered and displayed on different machines, in parallel, or it can be generated rendered and delivered within a single process.
• Run time data size checks determine where the data is rendered.
• It will display multiple views locally or in a cave or on tile display walls.
• During camera manipulation, the data and image may be down-sampled to maintain interactivity or full resolution images may be generated.
• It supports offscreen rendering and offline movie generation.
• It can use Mesa or Manta for pure software rendering but will take advantage of advanced GPU features (for depth peeling for example) when a graphics card supports them.
• The rendering architecture is extensible via plugins.

ParaView's display architecture is necessarily complex to support all of these requirements. As you create VTK pipelines on the server, ParaView creates one or more parallel and distributed memory split display pipeline for every VTK filter along that pipeline. These display pipelines manipulate for, deliver to and render the data within every window that shows the data.

The architecture of ParaView's display for ParaView 2.x is explained in: The ParaView user's guide version 1.8 Chapter 22.

The architecture of ParaView's display for ParaView 3.0-3.8 is diagrammed here:

Media:Pv_pipelines.pdf

The architecture of ParaView's display for ParaView 3.10 and beyond is diagrammed here:

Media:Pv_pipelines2.pdf