-an approximate number, as not all bits in
a 16-bit image are used to define color).
Each pixel in an image represents a single color
resulting from the computation of the underlying bits of data. Each
individual pixel in an 8-bit color image can be any one of 16.7
million colors.
Bit Depth
The terms "Bit Depth" and "Color Depth" refer to the number of
bits used to store an image, typically 8 or 16. An 8-bit color image
might be referred to as either 8-bit or 24-bit (24-bit equals three,
8-bit grayscale bytes or channels). Both are generally accepted to mean the same thing.
The same is true of 16-bit files which might also be referred to as
48 bit files.
Some devices, like scanners, have a bit depth of 10 or 12 but
produce a final file that is condensed down to 8-bit. Scanning at a
higher bit depth results in greater accuracy, even though the
information is condensed to 8-bits following the scan. There also
are 12 and 14-bit scanners that produce 16-bit files.
Many digital cameras provide the option of capturing
and saving
images in 8- or 16-bit mode and there is a case to be made that
16-bit capture provides advantages even if the file ultimately ends
up in 8-bit at some future point. While it might seem obvious that
16-bit mode captures color more accurately, there are some
practical reasons why a 16-bit workflow isn't necessarily the best choice. At
the same time, 16-bit has a rightful place in the digital workflow,
because
there are situations where the additional data provided by
16-bits offer advantages over 8. But first, a few practical
considerations:
16-Bit Workflow Considerations
It is generally accepted that human vision is capable of
distinguishing, at maximum, about 10 million colors. An 8-bit image
has a range of 16.7 million colors, clearly in excess of what
humans can distinguish. So while 16-bits' 281 trillion colors sound
impressive, they are far beyond what can be seen by the human eye. Additionally, most
photographs contain only a fraction of possible colors, perhaps a
few thousand to a few hundred thousand, well within the range of
8-bits.
File size doubles between 8-bit and 16-bit. An 8x10"/300ppi image
in 8-bit is 20.6MB in size, in 16-bit it's 41.2MB. Obviously, the
larger the file, the more room it takes for storage and the longer
it takes to transfer the image from one device or location to another.
Depending on the speed of the computer and size of the image
files, 16-bit files can take considerably longer to edit using
Photoshop or another image editing program. Additionally, some
editing functions, filters, and plug-ins are not available in 16-bit.
The vast majority of photographic output from the silver-based
prints we make at DigiGraphics to both home and commercial inkjet
printers, as well as displays (monitors) operate in 8-bit. When a 16-bit file is encountered, it
must first be converted to 8-bit within either Photoshop or the
behind the scenes within the
printer's software.
The 16-Bit Advantage
The strongest case for using 16-bit images deals with what
happens during image editing in Photoshop. Image editing, and in
particular setting curves to adjust contrast, can result in unwanted
banding or posterization in certain parts of an image. The areas
most susceptible to this are gentle color gradients like that found
in a blue sky. Non-photographic computer generated images that
contain gradients are highly prone to unwanted color banding.
Editing an image in 16-bit can all but eliminate banding. Once
editing is compete, the image can be successfully converted back to
8-bit for printing or other use.
Another consideration deals with image manipulation. Editing an
image changes the numerical values associated with each changed
pixel. Due to rounding, some degree of error is introduced in
calculating the new value. Working in 16-bit reduces the impact of
rounding errors. Images that will be subjected to heavy editing are
considered good candidates for 16-bit.
Like so many things in digital imaging, there is robust
discussion and disagreement within the professional community about
the overall benefits of working in 16-bit color. There are those who
swear by 16-bit and scoff at 8-bit workflows, and those who see
virtually no benefit over an 8-bit workflow.
While there is no disagreement that 16-bit images are
mathematically superior to 8-bit, there is considerable disagreement
as to whether or not 16-bit is visually superior. The fact that the
existence of visual differences is argued about provides some evidence about how finely the
hairs are split between 8 and 16 bit.
I belong to a professional imaging group named ColorTheory, with members
representing top-notch commercial imaging experts and retouchers from around the world. A standing
challenge was issued several years ago to anyone who could
demonstrate a real-world image (not computer generated or
unrealistically tweaked) in 16-bit that was superior to the same
image in 8-bit. As of today, no one has been successful in
demonstrating 16-bit superiority... although several continue to make
the claim without proof.
I am of the mind that if a 'benefit' isn't clearly detectable, it
shouldn't be
employed as a standard workflow. In situations where we know 16-bit
offers a visual advantage, as in dealing with gradients and/or special effects, it is
rightly employed because it solves problems that
8-bit can't. Working in 16-bit is less about rendering a color that
can't be found in 8-bit; 8-bit already provides more colors than we
can see. It's place is in certain editing situations where
8-bit falls short.
Future Advantages of 16-Bit
My crystal ball is a little fuzzy as to what the future may hold
for 16-bit. I suppose it's possible that images that have been shot
and saved in 16-bit will benefit from some future technology, and
that may provide a marginally valid reason to shoot in 16-bit until that
time comes. My guess however, is that greater gains will be made to
image quality through improvements to camera sensors than will ever be
made to improvements in 16-bit workflows.
Anecdotally, I edited perhaps 10,000 images over the past year and recall
only one that required editing in 16-bit (to solve a banding problem
in a gradient). Based on my experience, I recommend shooting and
editing in 8-bit unless there is a reason to do otherwise.
If
your experience is different, and you've found that working in
16-bit provides observable visual benefits not available in 8-bit, then by all means
continue to use the workflow that best suits you. I might add
that my retouching group would be interested in seeing comparative
samples that clearly show 16-bit to 8-bit superiority. If anyone sends
comparative
samples to me, I'll post them for review and discussion
among the group.
Lastly, because the equipment we use at
DigiGraphics cannot print 16-bit files directly, we discourage
submitting 16-bit files for output. If you've chosen a 16-bit workflow for your
own work, it is advisable that you convert your images to 8-bit,
making any additional adjustments to the image that may be necessary
following the conversion, before they are submitted for printing or
other service.
Additional Reading
Additional articles and differing opinions about 8-bit and 16-bit
workflows:
http://www.steves-digicams.com/techcorner/December_2006.html
http://www.photoshopessentials.com/essentials/16-bit/
http://www.earthboundlight.com/phototips/8bit-versus-16bit-difference.html
http://www.luminous-landscape.com/forum/lofiversion/index.php/t9785.html
http://www.northlight-images.co.uk/article_pages/16_bit_black_and_white.html
