RGB 48bits for "raw image data" loader (and other similar improvements)
I would like to see 16bits per channel (48bits per pixel) RGB and BGR images (in both little and big endian formats, and for both signed and unsigned values) get supported, in the same way that 16bit Gray images are now supported. Same thing for RGBA 64bit images, but now there also comes a question of where the alpha channel is located, as in RGBA, ARGB, BGRA, and ABGR. By the way, I notice that your 32bit RGBA (8 bits per channel) and 24bit RGB only supports one channel arrangement.
So I would like to see you support 32bit RGBA, ARGB, BGRA, and ABGR as well. And also on that topic, your 24bit RGB, should also allow the BGR version.
In total the list of available RGB types of raw images (not counting those with fewer than 8-bits per channel such as 16 bit RGB565) should include:
24bit RGB, 24bit BGR
32bit RGBA, 32bit BGRA, 32bit ARGB, 32bit ABGR,
48bit RGB (little-endian signed), 48bit RGB (little-endian unsigned), 48bit RGB (big-endian signed), 48bit RGB (big-endian unsigned)
48bit BGR (little-endian signed), 48bit BGR (little-endian unsigned), 48bit BGR (big-endian signed), 48bit BGR (big-endian unsigned)
64bit RGBA (little-endian signed), 64bit RGBA (little-endian unsigned), 64bit RGBA (big-endian signed), 64bit RGBA (big-endian unsigned)
64bit BGRA (little-endian signed), 64bit BGRA (little-endian unsigned), 64bit BGRA (big-endian signed), 64bit BGRA (big-endian unsigned)
64bit ARGB (little-endian signed), 64bit ARGB (little-endian unsigned), 64bit ARGB (big-endian signed), 64bit ARGB (big-endian unsigned)
64bit ABGR (little-endian signed), 64bit ABGR (little-endian unsigned), 64bit ABGR (big-endian signed), 64bit ABGR (big-endian unsigned)