It doesn't have 24-bit sensitivity but 10-bit, hence 10MB instead of 24MB. The pixels on the sensor can't detect R G and B on their own, they can distinguish just one color (R G or B) in 10-bit sensitivity. After shooting the picture, the 24-bit RGB is interpolated from the each pixel and its neighbours. So, the 10MB you're getting is really the complete RAW buffer with the RAW data from the sensor itself, before any postprocessing (except for badpixel mapping and possibly noise reduction).
More information:
Charge-coupled device - Wikipedia, the free encyclopedia
Bayer filter - Wikipedia, the free encyclopedia
Wow! I didn't know that!
I always thought that each of the 8Mpixels is a full fledged RGB pixel...
That means that my true "RGB" pixel resolution is around 2MPixels with 40bit depth for each pixel. (According to Bayer_filter of 2x2 - Thanks for the link!)
That explains why all images i shot at maximum resolution looks soft when 100% cropped, right?
It would look sharp if i half the resolution...
So, if i get it right, the JPG at 8Mpixels inflates and extrapolates the picture to "create" 24bit depth 8Mpixels,
while in fact the true resolution for 100% should be lower. (But keeping the same file size)
And one more thing,
10bit is considered good compared to other Point-And-Shoot cameras?
I know that SLR's come with A/D's of 12 and even 14bits so it's interesting to compare.
I was reading the review of the G9 at
Digital Camera Reviews and News: Digital Photography Review: Forums, Glossary,
FAQ and it states that the size of the RAW images is 12.5Mbytes,
which gives for the G9's 12Mpixel resolution a bit-depth of approximately only 8bit per pixel! Could it be?
