What made it faster? Is it just that the hardware could be simpler? Curious to know.
You could compose out of, say, 3 bpp sprites or layers. Or you could make 1bpp "control" bitplanes that, when fed to the blitter, would fill your polygons scanline by scanline. (The blitter can also make these for you using the harware line drawing support) All the graphics hardware could be bit depth agnostic but still do 1bpp operations 5x as fast as 5bpp operations. And do it asynchronously with CPU doing other stuff concurrently. Saved memory bandwidth, memory space, and CPU time - all of which were very tight and CPU had no cache.
Related note, there also used to be bit-sliced CPUs that didn't have hardwired byte lengths.
On the PC, VGA had a "byte-planar" mode that combines some of the advantages of planar and chunky graphics that had further advantages in eg filled polygon rendering over "plain" chunky mode.
This was also Amiga downfall, as planar graphic modes became irrelevant in the nineties. As soon as you get into ~486DX2/66 territory you have enough pure CPU to render chunky(13h) mode in software.