Swizzling in the context of computer graphics is the rearrangement of vector information in the pixel of an image in order to display it quickly and more efficiently. This is because the swizzling is done in favor of the way that a specific graphics processing unit (GPU) or CPU reads the pixel information, which reduces the overhead. The result is the same image that can be rendered much faster.
Swizzling is achieved by rearranging the matrix of values that make up the vector information. This happens because different renderers and graphics processors access data in different ways and in different orders, so swizzling vector information in that particular way allows the programmer to use specific instructions that can process data more quickly. This can also help organize the data in memory to make better use of texture caches. For example, the vector component c1.rgba can be moved to c2.abgr to remove a step it takes a graphics processor to reorder the data, thereby speeding up processing.
A disadvantage of swizzling is that it makes quick access to the image data difficult because it has changed dramatically from its original shape. A Swizzled image could therefore be processed more quickly, but is very difficult to edit afterwards.