Thomas T. Veldhouse wrote:
Scott W wrote:
You are missing the big reason, 1 color / pixel for raw vs. 3 colors /
pixel for tiff.


True ... but there are Red, Green and Blue pixels, each getting 12-bits of
linear data, which makes up one pixel [three colors] in TIFF. In fact, for
some reason, I believe Bayer sensors have two green pixels for every RGB pixel
in TIFF.


It's interpolated. In reality, if you have an 8000000 pixel camera,
each of these pixels receives either red, green or blue light. Thus, 12
bits per pixel in the raw data (each pixel records one colour). What
conversion from raw to eg tiff does is, basically, to take these and
interpolate so that each pixel is assigned a colour (not just a value
for r, g or b, but one for each). Thus, an 8mp raw uncompressed file
will be around 12MB, while an 8mp 8-bit (per channel, as 8-bit usually
signifies) tiff will be around 24MB (and 48MB if 16-bit).

If you don't believe me, google for IRIS, download it and look at a raw
file. Or google for demosaicing, I'm sure you'll get lots of hits.

Whether the format is linear or not is completely irrelevant to the
size of the files; it only has to do with their interpretation.