# 8bit and 16 bit channel colour

Can anyone explain what it means to have an RGB mode with 16 bits per channel? I understand that 8 bits can give 256 different values per channel, and with three channels, you get 256 x 256 x 256 = 16.7 million colours. But why would one need 16 bits per channel? 16 bits per channel means, to me, 65536 values per pixel. With 3 channels, this means 65536 x 65536 x 65536 or several trillion possible combinations per pixel, which is unthinkable.

Can anyone also explain why, when saving a photoshop document as a .bmp file, why do you get the options to save it as 16 bit or 32 bit? I thought that a .bmp could not be more than 24 bit (3 x 8 bits, or three bytes, one byte for each colour 0-255). In the 32 bit mode, what is the extra 8 bits for? I understand why you can save as 8 bit (grayscale or indexed colour) and 1 bit (line art). Also, 4 bit is confusing, What's that for?
LVL 11
###### Who is Participating?

x

Commented:
Hi Jason

Sorry, my mistake, I think you right, it's 8bit per channel and not 8bit for all 3 channels. (http://www.nebulus.org/tutorials/2d/photoshop/color/ ) - (quite alot of info on colours).

So for an 8 bit colour channel it mean 256 different shades of that colour. and like you said for RGB it will be 256*256*256 combinations of colous of each shade. So regarding your question about 32bit images - as I said in my previous post, I don't think it for the human eye, as we cant really tell the difference between a 24bit colour image and a 32bit colour image. So it has no use for our eyes, but maybe for machines, that need to go into fine detail for analysis or huge poster printings, high bit colour depth becomes useful.

regards-
0

Commented:
Hi

I don't think the calculation of 8 bits colours will give you 256 x 256 x 256 = 16.7 million colours.
It works more like 2^8 = 256 colours, you can only get 16 million when you have 24bit colours.
Something like this:

4 bit color  16 colors (2^4)
8 bit color  256 colors (2^8)
8 bit grayscale  256 shades of Grey (2^8)
24 bit color  16 million colors* (2^24)
30 bit color  1 billion colors* (2^30)
36 bit color  Billions of colors* (2^36)

Our human eye, can't really tell the difference once it reaches true color (i.e. 24 bits), i think it more for machine use, such as image manipulation, analysis etc... rather than for the human eye. If it just for viewing by human increase in colour depth will be just a waste of storage space.

Here are some references that you can have look at.

http://fla.esf.edu/research/CAVLab/reference/candbd.htm
http://www.luminous-landscape.com/tutorials/bit-depth.shtml
http://wonders.eburg.wednet.edu/Topics/GMMIE/colorDepth.htm

regards-
0

Author Commented:
>>I don't think the calculation of 8 bits colours will give you 256 x 256 x 256 = 16.7 million colours.
It works more like 2^8 = 256 colours, you can only get 16 million when you have 24bit colours.

What I meant was you can have 3 8-bit channels to define a pixel, which is t2^8 x 2^8 x 2^8 = 16.7 million.
0

Author Commented:
that makes sense thanks.
0

Commented:
Often the higher bit-depths are useful if there is to be any image manipulation afterwards.  For example, cleaning up colors by adjusting gamma can result in losing colors, so if you start with more than you need, the image will still be in fine shape afterwards.  Check out this article:

http://reviews.cnet.com/4520-6501_7-5334452-1.html?tag=dir

0

Commented:
32 bit BMP is for 4 channel RGBA BMP images which have an additional 8 bit alpha channel (transparency) apart from the 24 bit color information. This is more or less suported directly by Windows XP and partly by Windows 2000. Several newer XP icons come in 32 bit versions. 4 Bit BMP is a image with only 16 colors. This was used quite often by early versions of Windows, even some XP icons still come with only 16 colors. 16 bits per channel is rather uncommon and not possible with BMP but is supported i.e. by PNG. Photoshop and some other programs are able to use 10 bit per channel internally which results in 30 bit color information per pixel or 40 bit with alpha channel.
0

Commented:
> But why would one need 16 bits per channel?

Sometimes an highe bit-depth is necessary to store scientific data: graphical representation of the image is good for qualitative analysys, while numerical data are good for analysis.

0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.