Terminology

We use the following terminology to talk about the way how SGCT works.

There is a single Cluster that consists of 1 or more Nodes. Each Node usually corresponding to a single computer, but multiple nodes can be executed from the same computer, although this is rarely the most optimal solution. Usually each Node will be an individual computer. All Nodes should be connected via low-latency network connections. One of the Nodes in the Cluster is designated as the server, whereas the other Nodes are clients or Image Generators (IGs). Even if a Cluster only consists of a single Node, it is still considered a Server without any clients.

Each Node contains 1 or more Windows. Each window is using the operating systems default representation and its resolution, position, and other parameter can be configured. The resolution and position are defined in the native units of the operating system, often logical pixels.

Each each Window containing 1 or more Viewports. A Viewport is a part of a windows into which some 2D or 3D content can be rendered. The position and the size of each Viewport is described in a 0-1 range.

If a Viewport is set to render 3D content, it must have one of many Projection types that determine how the 3D content is rendered. The most common projection type is the PlanarProjection in which a horizontal and vertical field-of-view is defined. Another often used projection type is the FisheyeProjection in which multiple render passes are automatically combined to generated a square fisheye or dome master image.