Let me try...
Each sensor pixel records one 12 bit color value. Software combines the info from a 4 pixel clump set up something like
RG GB
Where the green pixels are also used to provide luminance (brightness) information. (I have never quite understood whether luminance is just taken from the second green pixel, as a combined value.)
That software takes the combined color and luminance value and reassigns it as 4 separate pixels (Good software weights these values by the values from the adjacent clumps), and writes them as 12bit values padded out to 16bits per pixel (because the kind of computers we use have a module 8bit requirement), or drops the lowest order 4 bits and writes it out as 8bits per pixel.
A RAW file is just the actual sensor pixel data padded to 16bits per pixel (4 meaningless bits added). Software has to do the stuff mentioned above to it either in the camera, or in your computer, to convert that data into a picture.
, to answer Shel's question, that interpolation is based upon 12bits per pixel no matter whether the final output is chopped to 8bits, or padded to 16bits.
--
Rob Studdert wrote:
On 12 Sep 2004 at 22:14, Shel Belinkoff wrote:
I asked about 12-bit color space. That was my terminology.
OK, not quite the correct terminology, a colour space can be defined independent of bit depth, see: http://www.drycreekphoto.com/Learn/color_spaces.htm
But, if I understand what you're saying, then the istd does not record 16
bits/channel, rather only12 bits. The 16 bits comes about thru some sort
of interpolation or mathematical wizardry. Perhaps this is an
oversimplification, but that's ok for me for now. And you say that this
interpolation will "provide up to 16 bits precision." That implies that it may
not actually provide 16bits, only up to 16bits in certain circumstances.
It's all wizardry, all outputs except RAW (which isn't a conventional image format as such) have to be interpolated. The *ist D sensor has 6.3MPixels of which 1/2 record green luminance, 1/4 red and the remainder blue yet at any bit depth (8 or 16) output it still delivers 6.1MPixels with a red green and blue luminance value at each pixel. This is true for any digital camera that has a single sensor and isn't a Foveon design.
The Pentax RAW convertor will provide TIFF files at 8 or 16bit colour depth, and the PS CS convertor will decode and import a Pentax RAW file as an 8 bit or 16bit (per colour) image. Regardless the RAW file always contains 12 pits per pixel (an 8 bit JPG file contains 24 bits per pixel)
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
-- graywolf http://graywolfphoto.com/graywolf.html
