This is NOT suppose to be a complete explanation of everything related to bit depth and channels, it's only an explanation of the bit depths in photoshop in relation to each other!
Remember, 8 bit, 16 bit, and 32 bit images are NUMBER OF BITS PER CHANNEL! There are basically 3 channels in an RGB image, so that's like 24 bit, 48 bit, 96 bit respectively. That is because one term describes the number of bits per channel, while the other describes the number of bits per pixel. 32 bit often refers to 24 bit, though 32 bit is actually 8 bits per channel, with an extra "alpha" channel (for transparency).
Notes: I based the 16 bit, which is kind of the key here, on experience in Photoshop, 16 bit might be different in other applications, and for other file formats (Like TIFF I think).
Also, I call RAW 12 bit, that is because in most cases RAW image files contain 12 bits of information per channel. RAW differs between camera manufacturers and camera models, so they work in very different ways. Some RAW files are 10 bit, or 14 bit. No "ordinary" camera can output more than 14 bits though.


































-- Why is the 16 bits / channel bar graph the same size as the 8 bits / channel graph? (and the 12 bits / channel graph much bigger than both?)
-- You should emphasize the difference between total color resolution and bits / channel. The designations of 8 bit, 16 bit, 24 bit and 32 bit color have been used for a long time, and all refer to 8 bits / channel (color) RGB (plus an 8 bit alpha channel for 32 bit color.)
-- Bits / channel is only part of the story. What about color space? --RGB (3 channels), CMYC (4 channels), LAB (3 channels)....etc.
-- Raw isn't limited to 12 bits / channel. Raw is just .... raw data. Some digital cameras have 14 bits / channel. And a 12 bit resolution image can be expanded (or compacted) to fit any channel--sure, it doesn't contain any more real data, but there isn't much point of converting to a 16 bit channel without that expansion (compacting (12 -> 8 bits) would discard data....) So no contrast is lost, but the full range of intermediate gradation isn't utilized on a larger channel.
-- Anytime you add a masking layer, you add another channel (an alpha channel.)
This stuff isn't too complex:
-- Bits per channel == # of bits used to represent each component in a color space (RGB or other color space)
-- Total colors == highest # that can be represented by that # of bits (unless it's floating point, which is a different animal.)
But it looks like you clarified much of the text, anyway.
I'd like to comment in more depth, but I gotta go photograph some chump who runs one of those corporate touchy-feely seminars...
gamer: what?
randofo: This stuff confuses the heck out of me.
But wow, that looks really complicated, amazing how you know this... I would probably go insane trying to do this kind of stuff.
Nice job! :P