“Shadow” is intended to imply that each pair of cards has a primary image and a “shadow” which in its original form represented additional associate images, but can also mean that each suit pair has a primary combo and a related shadow combo, which results in a 2-block (1352 image, two pairs per image) system.
Shadow’s suggested 2-block approach is to use variable image stacking at each loci. A red-first pair signals that you should end the scene at the loci amd move to the next. When you recall your scenes, the last element at each loci represents a red-first pair. There is no mention of multiple palaces, nor is it suggested to use them.
There is a technique that some have called “variable spatial encoding” which involves using placement at the loci to determine the color value, but it is generally not as effective, especially when going for speed.
One palace, one route, variable number of images per scene depending on color. That is how Shadow was built.