On 7 Sep 2004 at 12:42, Stuart Moore wrote:

> I spent some time on the Microsoft website. As far as I can work out, sRGB is
> the built in default and indeed it is "based" on monitor capabilities but
> Windows is colour space neutral in that it will use any colour space properly
> provided and there are published APIs (amongst other things) for this. However,
> the market for alternative colour spaces is small and specialist so there are
> only a few alternatives out there and they tend to be expensive. 

Most current monitor "drivers" produced to suite late Windows variants 
(98,XP,2K etc) will load a monitor profile under the "colour management" 
display tab however its validity is limited by the set-up of the monitor. 
Working colour space profiles are generally distributed freely. Specialized 
output device or custom profiles for consumer printers/scanners sometimes cost 
an arm and a leg however custom profiles supplied by a pro-lab set-up to 
provide optimal results on their equipment is usually supplied free.

> A number of
> programs including Photoshop bypass the Windows management and implement their
> own colour space (not a huge risk for the market leader in graphics to take as
> everyone else will make their products work with it anyway regardless of
> Microsoft standards).

Photoshop introduced the gamma tool which was initially set-up to implement 
some colour calibration to the display system. It modifies the display card 
look-up table (LUT) to normalize gamma however the gamma tool can be discarded 
if you are using a calibrated monitor as Photoshop always utilizes the Windows 
monitor profile, you can however set the default working colour space to 
whatever icc profile you choose.

> Looking through various postings about the latest PSP, my view of the consensus
> is that JASC decided not to try and hack Windows themselves or license an
> existing alternative colour space model on the basis that there is not much
> demand for it from the target customer base (those with more sophisticated
> requirements assumed to be using Adobe Photoshop I guess).

Even ThumbsPlus (which is primarily a image file management program) provides 
the option to "Obtain current monitor profile from Windows on program startup" 
and also a default working space (used to render images which do not have an 
embedded CS). It's a long time since I used PSP but I'm surprised that it 
appears they've not yet reasonably implemented colour management.

> That aside, it is not clear to be what I am losing out on as the RAW images from
> the *ist D seem to be only 12 bit (16 bit with 4 bits zeroed) let alone 24 bit.
> What might I want to do to the images that would only work with increased bit
> depth?

Not sure what you are asking here? The sensor system delivers 12bits per pixel 
RAW in a matrix of RGBG. You must consider that for each decoded output pixel 
at 1:1 scaling a cluster of RAW pixels are used to interpolate the final pixel 
value. The precision of that calculation is limited to 16 bits per colour 
channel (48bit colour) but can be truncated to 8bits per colour channel (24bit 
colour). Obviously there is a limit in the usable bit depth which would be a 
function of the decoding algorithm employed plus the system noise etc but I'm 
definitely sure it would be reasonable to assume at least 14bits per colour 
channel of usable output precision and even if the last bits are noise 
components they will act as a dither of sorts.

If you generate a 24bit and 48bit TIFF from a single RAW PEF file you will note 
the colour count is far higher for the 48bit image (PSP used to have a colour 
count option). An image generated with 16bits per colour channel can be 
manipulated by a greater degree without incurring the penalty of visible 
artifacts such as banding.


Rob Studdert
HURSTVILLE AUSTRALIA
Tel +61-2-9554-4110
UTC(GMT)  +10 Hours
[EMAIL PROTECTED]
http://members.ozemail.com.au/~distudio/publications/
Pentax user since 1986, PDMLer since 1998

Reply via email to