RAW, 8 Bit, 16 Bit, and 32 Bit Explained





Introduction: RAW, 8 Bit, 16 Bit, and 32 Bit Explained

After alot of confusion about the 8bits/channel, 16bits/channel, 32bits/channel, and RAW color modes in photoshop, I finally figured it out, and decided to make an illustration to show how I've understood it.

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.



    • Microcontroller Contest

      Microcontroller Contest
    • Science of Cooking

      Science of Cooking
    • Pocket-Sized Contest

      Pocket-Sized Contest

    We have a be nice policy.
    Please be positive and constructive.




    I think you've confused as many people as you've enlightened....

    -- 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 bitcolor 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.)


    The whole point is that the 16 bit color space has the same range as the 8 bit, but with higher resolution per tone... that's why it's the same size, because white in 8 bit is white in 16 bit. In 12 bit you can change the exposure. Mean that what is white at one exposure might be white in a higher exposure, or it might be grey. All depends on exposure, like when taking pictures. This isn't suppose to be a complete explanation on color spaces, this is just to illustrate the differences between the bit depths you encounter in photoshop (Which should be pretty obvious) The only reason I included the 8/16/32bit-24/48/96bit explanation was because I found that alot of people, when discussing these color depth topics often get confused. Raw is generally 12 or 14 bit, at least for digital cameras (like, you know, the ones that spit out the version of raw data we work with here...) BUt yea, I forgot the note that RAW isn't fixed to 12 bit, but can also be 14, or 10 bit. And I don't see how masking(/alpha) has anything to do with this either, because as I said, this is not "the basic guide to bit depth and channels".

    At least include the word "channel" in the title, since (as I said), the terms 8/16/32 bit have been used to describe color spaces/total color res for 12 - 15 years....That alone is cause for much confusion.

    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...


    Well congratulations. And hope my explanation of RAW tone range vs. 16 bit (per channel) tone range could help you a bit with your photography hobby, or whatever it is...

    Sorry. I've been a professional photographer for over 20 years (with a degree in photo illustration.) And I've also taught photography at both the high sch (private school) and the college level (including Photoshop.) But a resume isn't a valid argument (neither is ignorance.) Personally, I don't care for your tone. Since I don't have anything nice to say, I'm done here....


    Nice to know you're finally done, because to be honest, I think you brought some good C&C; to my slide, but your tone really sucks. It seams like you think you can smear on a fat layer of ignorance because of your experience, and that pisses me off, because you don't even seam to care that I put both time and efford in creating this illustration so other people can avoid the confusion I've had. So yea, I might have been answering you in a harsh tone, but that's because it's the tone you deserve to be answered in...

    Yes, no more comments about this (although my original comments still stand, and are accurate.) But good luck anyways....if you choose photography as a profession.

    Correction: only 24 bit color and 32 bit color are 8 bits / channel. 8 bit color and 16 bit are lower (usually indexed) color formats.


    Where does this correction apply? Please note that it's all "Bit PER CHANNEL"...