Some Chips support 12bit per color element (total of 36 bits per pixel) which are not byte aligned and thus the current Feature/Method model can't support.
The feature model today assumes a pixel of data is always in byte length sizes.

Describe the solution you'd like
Enhance the features to support bits rather than bytes for pixel length.
Some methods use look up tables to quickly encode/form the output data for the hardware peripheral they use; these tables also makes assumptions about the size, but often it is only two bits minimum.

Describe alternatives you've considered
Two or four bits size increments due to method limitations.

Additional context
#705