Why modeling image's pixel with char instead of unsigned int?

Question

I study different image processing project (in C\\C++) to help me build a ColorGrading App. Including OpenColorIO, it seems that using char/unsigned char i