RAW+JPEG WHERE??? HOW???

That's very true but for all practical intents (disregarding the internal compression mechanisms) JPEG offers 24bit (8bits/cc) colour to the user when opened it's just that the RGB values are interpolated rather than directly stored as is the case with non-destructive image formats.

Pixel is a very general term, pixels on a capture device (excepting Foveon sensors) record a single primary colour each. Display devices output a single colour per screen pixel (sometimes referred to as sub-pixels) though the input file may contain composite RGB information per pixel. A pixel in a normal image file generally consists of a trio of RGB values.

Basically in the case of image files if R,G and B channels each contain 8bits of information the pixel will contain 24bits. When working with 16bit files each pixel contains 48bits, 16 bits for each colour channel. I should add that certain file types can also accommodate additional non-RGB layers (such as TIFF and PSD) for masking or other specialist applications.

RAW files from a Bayer array sensor generally represent a single colour per pixel and each pixel contains between 12 and 16 bits. Only after the RGBG Bayer matrix has been de-mosaiced by the RAW converter will a composite file with RGB per pixel be available.

The following links aren't the best I've read but may serve to provide a little more information:

http://en.wikipedia.org/wiki/Pixel
http://en.wikipedia.org/wiki/Bayer_filter
http://en.wikipedia.org/wiki/Demosaicing

Perhaps the term "sub pixel" should be used - to mean a single-colour unit.

Well, yes... as an image format that's fine. But, I understood we're talking in the context of image capture, so implying that "RAW contains 12-bits per pixel, but JPEG contains 24-bits per pixel" is misleading

After demosaicing, the RAW actually contains 36-bits per pixel (12 each of R, G and B) (or 42 in the case of the K20D). (And the JPEG from camera was demosaic'd anyway from the RAW).

As far as luminance is concerned, a JPEG contains 8-bits per pixel whereas RAW contains 12-bits per pixel. (That's effectively the black and white content for anyone that's wondering ) Typically you find the limits of JPEG when trying to bring out detail in shadow areas and the like.


And now I suspect everyone is confused
Matt

That's why I said JPEG are 24bits or 8bits/cc, I don't believe that it's misleading in any way. Regular JPEG files are encoded from a 24bit file and decoded to a 24bit file, yes?

I understand that the K20D RAW files are 12bits/pixel the same as the K10D, the ADCs have differing sample bit depths but the RAW files for both still remain at 12bits/pixel after limited in camera RAW file gain control. Again it's an oversimplification to say that the de-mosaiced RAW file contains a discrete 36bits, it's interpolated using floating point math so unless it's been truncated to 8bits/cc on import it may actually contain more than 12bit/cc of usable data (considering the colour balancing and gamma curves that the RAW data has been subjected to). Let's face it a huge percentage of the data in a de-mosaiced image is interpolated.

LOL, yes that's correct in a very RAW interpretation.

HELP!!!!
CHEERS Vic

Thoroughly

(but don't mind me - the camera could be full of elves with paintbrushes - as long as they continue painting, I'm happy... )