HP SVA V2.1 Parallel Compositing Reference Guide

Output
An output is a subrectangle of a frame that a host needs to retrieve once the frame has been
produced. A host declares at most one output that it needs to retrieve. The dimension and location
of this output on the frame are stored in the context. Two different hosts can declare the same
or different outputs.
Result
A result is a sub-rectangle of an output. The host can retrieve its output either as a whole or in
pieces. Each rectangle of pixels retrieved from the output is called a result. Based on the specific
operator being used, a result has one or more channels of data. For example, a result associated
with the depth operator has two channels:
• Color channel: composited pixels.
• Depth channel: corresponding depth value of each pixel.
A host can retrieve zero or more results from the output it has declared in the context.
Pixel Operator
The Library combines, that is, composites pixels from multiple framelets. Different algorithms
are available to do the compositing. These algorithms typically take two pixels as input and
produce a single pixel as an output, similar to an arithmetic operator. Consequently, each of the
algorithms is referred to as a pixel operator. See Section and Table 2-10 (page 33) for more
information.
How Library Entities Are Related
Figure 1-1 illustrates the relationships among the entities defined by the Library. The clouds
name the entities. The lines state relationships among objects. As illustrated in the figure Key,
the relationships are of three types:
• Ownership (black dot)
The entity with the dot owns the entity on the other end.
• Use (white dot)
The entity with the dot uses the entity on the other end.
• Relates (no dot)
The two entities are related by the verb that labels the relationship. Such a relationship is
read from left to right (for example, one or more hosts share a context).
How Library Entities Are Related 13