Apple Shake 4 - Manual (Page 764)

764 Chapter 26 Transformations, Motion Blur, and AutoAlign
Concatenation of Transformations
Many of the transform nodes concatenate, similar to the way color-correction nodes
concatenate. Like color corrections, compatible transform nodes that are connected to
one another are concatenated so that each operation is collapsed into a single
calculation, optimizing both processing time and image quality. You can tell which
transform nodes concatenate by the letter “C” in the upper-left corner of the icon in the
Transform tab. In the screenshot below, you can see that the CameraShake, CornerPin,
and Move3D nodes all concatenate, but the Orient node does not.
Thanks to concatenation, you can apply a Move2D node, a Rotate node, and a CornerPin
node, and Shake resolves all three nodes by collapsing them into a single internal
calculation, thus executing the end result in one operation. Not only is the render
optimized, but the resulting image is of much higher quality (since the image is filtered
one time instead of three times).
For example, if you apply a Rotate node to an image of the earth so that the lit side
faces the sun, and then use a second Rotate node to orbit the earth around the sun,
Shake determines the combined transformation and calculates it in one pass. This is
important to understand because it means you spend (in this case) half of the
processing time, and only filter the image once.
Note: To see the destructive effects of repetitive filtering, try panning an image 20
times in another compositing application.
It is to your advantage to employ transform concatenation whenever possible. As with
color-correction nodes, transform nodes only concatenate when they’re connected
together, so arrange the operations in your node tree wisely. Any other type of node
inserted between other transform nodes breaks the concatenation.